[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7C6R5H2SW40BQ3GJ61SSFP4":3,"skill-same:01M7C6R5H2SW40BQ3GJ61SSFP4":44},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":14,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":29,"install":30,"resources":43},"01M7C6R5H2SW40BQ3GJ61SSFP4","start-feature",null,"Use when the user asks Pi to start a HEPHA FEAT implementation, run start-feature, or run start-implementing for a READY FEAT from a registered project workspace, for example \"use the start-feature skill for FEAT-003\".","aboimpinto\u002FAgenticDevelopmentProcess","aboimpinto","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F1231687?v=4","01M7C6P0WDWM5B2RBSKSPR66PZ","Paulo Aboim Pinto","01M7C6P0WXGYZP7QMFS91N8C3P",0,"MIT","https:\u002F\u002Fgithub.com\u002Faboimpinto\u002FAgenticDevelopmentProcess\u002Fblob\u002Fmaster\u002Fpi-packages\u002Fpi-skill-hepha-continue-implementation\u002Fskills\u002Fstart-feature\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: start-feature\ndescription: Use when the user asks Pi to start a HEPHA FEAT implementation, run start-feature, or run start-implementing for a READY FEAT from a registered project workspace, for example \"use the start-feature skill for FEAT-003\".\nagent_action: start-feature\n---\n\n# Start Feature\n\n## Model Authority\n\nThis procedure is model-neutral. When invoked directly, execution is\n`direct_host` and remains in the current Pi, Codex, or Claude Code session; that\nactive host owns model selection. Do not query Hepha routing policy, request a\nmodel switch, automatically hand off, choose a fallback model, or fabricate\nroute-policy evidence. Direct execution does not fabricate an orchestrated\nreceipt. Only an explicit Hepha launcher or dashboard dispatch creates a\nseparate `orchestrated` worker, whose route is injected outside this skill.\n\nYou are executing the HEPHA `start-implementing` workflow directly from Pi.\n\nThis skill is for console-driven implementation startup and for HEPHA WebApp\nworkers that mount the same skill. It must work when Pi is opened from a\nregistered project or parent workspace and the user names the active project\nplus a FEAT id such as `FEAT-003`.\n\n## Objective\n\nResolve the active HEPHA project and MemoryBank, locate the requested FEAT,\nperform the start-feature transition when the FEAT is still READY, then in\nautonomous mode continue contract-declared phases with verification, code\nreview, required fixes, and review reruns until every declared phase is\nresolved or a real blocker requires human input.\n\n## Required Inputs\n\nAccept these from the user's message:\n\n- project name or alias, such as `HEPHA`, `Hepha`, or\n  `AgenticDevelopmentProcess`;\n- FEAT id, such as `FEAT-003`;\n- optional MemoryBank path override;\n- optional mode, where `autonomous` means continue through subsequent phases\n  without stopping after the start transition.\n\nIf the FEAT id is missing, ask one concise question. Otherwise, infer safely.\n\n## Workspace And Project Resolution\n\n1. Treat the current directory as the workspace starting point.\n2. Resolve the active project named by the user and locate its repository root.\n3. Read the project `AGENTS.md` and project brief or MemoryBank overview when\n   present before editing.\n4. Resolve the active project:\n   - `HEPHA`, `Hepha`, and `AgenticDevelopmentProcess` map to\n     `\u003Cworkspace root>\u002FAgenticDevelopmentProcess`.\n   - If the user names another project, use `docs\u002Fprojects.md` or the nearest\n     child repo folder matching that name.\n5. Read the child project `AGENTS.md` and `README.md` when present.\n6. Run project git, build, test, and edit commands from the child project root,\n   not from the parent workspace.\n\n## MemoryBank And FEAT Resolution\n\nResolve the MemoryBank in this order:\n\n1. User-supplied MemoryBank path, if present.\n2. `.hepha\u002Fprojects.json` in the child project, matching project `name`, `id`,\n   or child root path.\n3. `\u003Cchild project root>\u002FMemoryBank` when it contains `Features\u002F`.\n4. A direct search under the child root for `MemoryBank\u002FFeatures`.\n\nThen resolve the FEAT:\n\n1. Normalize the requested FEAT id to uppercase, for example `FEAT-003`.\n2. Search these folders under `\u003CMemoryBank>\u002FFeatures`:\n   - `02_READY_TO_DEVELOP` first;\n   - `03_IN_PROGRESS` second, for HEPHA WebApp runs that already performed the\n     folder move or direct runs that are being resumed;\n   - then `01_SUBMITTED`, `04_COMPLETED`, and `05_CANCELLED` only for\n     diagnostics.\n3. Match a FEAT folder by folder name or by `FeatureDescription.md` content.\n4. Continue only when the FEAT is in `02_READY_TO_DEVELOP` or\n   `03_IN_PROGRESS`.\n5. Require `FeatureDescription.md`, `FeatureTasks.md`,\n   `PhaseExecutionContract.json`, and every contract-declared phase document.\n6. Stop if the FEAT contains `[NEEDS VALIDATION]`, and report exact file and\n   line references.\n\n## Start Transition\n\nWhen the FEAT is in `02_READY_TO_DEVELOP`:\n\n1. Inspect `git status --short --branch` from the project root.\n2. Stop before changing branches if unrelated dirty files make the transition\n   unsafe.\n3. Create or switch to branch `feat\u002F\u003CFEAT-folder-name>` unless HEPHA-provided\n   runtime context names a different target branch.\n4. Move the FEAT folder to `Features\u002F03_IN_PROGRESS`, preserving the folder\n   name and files.\n5. Update `FeatureDescription.md` and `FeatureTasks.md` so status and folder\n   references say the FEAT is in progress.\n6. Update linked EPIC references when the parent EPIC is available.\n7. Create or update `start-feature-report.md` in the FEAT folder with branch,\n   source folder, target folder, timestamp, and validation findings.\n\nWhen the FEAT is already in `03_IN_PROGRESS`, do not move it back to READY.\nRepair stale READY\u002FSUBMITTED references in FEAT documents, ensure a\n`start-feature-report.md` exists, then continue from the first unresolved\ncontract phase.\n\nCreate focused checkpoint commits at meaningful gates when safe, especially for\nthe start transition and review-approved phase gates. Never push during this\nskill; `complete-feature` handles pushes, merges, and completed-folder moves\nafter human gates are accepted.\n\n## Start-Feature Post-Operation: Estimates\n\nBefore any implementation phase begins, calculate and write the following for\nevery contract-declared phase:\n\n- `Estimated Human Time`: focused engineering time for a competent developer;\n- `Estimated AI Time`: active agent\u002Fruntime time, excluding waiting and human\n  checkpoints.\n\nUse only parseable compact values: `30m`, `1h`, or a same-unit range such as\n`2-3h`. In ranges, use the literal ASCII hyphen-minus (`-`, U+002D) only; never\nuse an en dash (`–`, U+2013), em dash (`—`, U+2014), or another typographic dash.\nBefore returning, inspect every phase estimate and the `## Implementation Timing\nSummary`; replace any typographic range dash with the ASCII hyphen-minus. Add\n`## Implementation Timing Summary` to `FeatureTasks.md`, with full-FEAT totals\nfor both estimates. Do not invent actual duration: HEPHA records it from completed\nphase worker start\u002Fend timestamps. Stop and report a blocker if any contract-declared\nphase lacks both parseable estimates or the timing summary before implementation starts.\n\n## Phase Resolution\n\nBuild the phase inventory before editing. `PhaseExecutionContract.json` is the\nonly source of phase topology and ordering; never infer it from a phase number,\ntitle, or filename:\n\n1. Read `FeatureTasks.md`.\n2. Read and validate `PhaseExecutionContract.json`, then read exactly its\n   declared documents.\n3. Parse phase status from the `FeatureTasks.md` contract inventory first, then from\n   a phase file status line such as `**Status:** COMPLETED`.\n4. Normalize statuses:\n   - `Completed`, `Complete` -> `COMPLETED`\n   - `Skipped`, `N\u002FA` -> `SKIPPED`\n   - `Not Started`, `NOT_STARTED`, `Pending` -> `PENDING`\n   - `In Progress` -> `IN_PROGRESS`\n   - `Checkpoint In Progress` -> `CHECKPOINT_IN_PROGRESS`\n   - `Code Review In Progress` -> `CODE_REVIEW_IN_PROGRESS`\n   - `Awaiting Review` -> `AWAITING_REVIEW`\n   - `Awaiting User Acceptance` -> `AWAITING_USER_ACCEPTANCE`\n   - `Blocked`, `Failed`, `Rejected` -> `BLOCKED`\n5. Stop with a blocker if the execution contract is missing, invalid, or its\n   declared documents do not match the `Phases\u002F` directory.\n6. The current phase is the first contract-ordered phase that is not `COMPLETED` or\n   `SKIPPED`, unless a review-finding report clearly requires returning to a\n   specific earlier phase.\n7. If every contract phase is `COMPLETED` or `SKIPPED`, scan each phase's\n   `## Quality Gate Evidence` table. If any gate decision is `missing`, treat\n   the first contract-ordered phase with a missing gate as the current phase for\n   quality-gate repair only.\n\n## Autonomous Phase Loop\n\nWhen the user says `autonomous`, continue through phases in execution-contract order:\n\n1. Start at the current unresolved phase.\n2. Complete or skip only that phase's scoped work.\n3. Update durable task evidence and narrative notes in the phase document. Hepha owns lifecycle\u002Fstatus fields and FeatureTasks status cells.\n4. Run focused verification.\n5. Run code review when needCodeReview is true.\n6. Fix blocking review findings, rerun focused verification, and rerun code\n   review.\n7. Record completion or skip evidence only after verification and review gates\n   are clear; Hepha deterministically writes the lifecycle state.\n8. Rescan phases from disk.\n9. Continue to the next unresolved contract phase.\n\nStop only when:\n\n- all declared contract phases are `COMPLETED` or `SKIPPED`;\n- all phase quality gates are `satisfied`, `waived`, or `not applicable`;\n- a real blocker requires human judgment;\n- credentials, permissions, missing dependency, or unsafe action prevents safe\n  progress;\n- the same failure repeats after a documented recovery attempt;\n- the user requested interactive mode and the start transition is complete.\n\n## Per-Phase Work Rules\n\nBefore changing code in a phase:\n\n- Read the phase document.\n- Read `planning-analysis-report.md` whenever the current contract phase's\n  planning-index row identifies it as required context.\n- Read relevant parent EPIC and FEAT acceptance criteria, including linked\n  `EpicAcceptanceTests.md` scenarios and the current phase's `EPIC Acceptance\n  Traceability` assignments in `FeatureTasks.md` when present.\n- Read applicable `MemoryBank\u002FLessonsLearned` documents.\n- Inspect current git status from the child project root.\n- Preserve existing checked Markdown checklist items.\n- For a V3 phase, `PhaseExecutionContract.json` is the machine task authority.\n  The `## Phase Task Ledger` must be its exact checkbox projection: one entry\n  per contract task, same order, matching `[contract:\u003Cid>]` and\n  `[executor:\u003Cexecutor>]` markers. Never add an uncontracted checkbox there.\n  Put descriptive work under `## Detailed Work` as plain bullets.\n- If the phase has no checkbox ledger, add a compact `## Phase Task Ledger`\n  from its declared contract tasks before substantive work.\n\nDuring implementation:\n\n- Work only the current phase scope.\n- Skip checked tasks unless a changed file, failed verification, or review\n  finding explicitly invalidates them.\n- Do not edit reserved machine fields: `**Status:**`, FeatureTasks Status cells, or Quality Gate Evidence Decision cells. Hepha writes them from durable workflow events.\n- Record task progress and evidence; Hepha applies the lifecycle transition.\n- Do not broaden scope to dashboard\u002FUI\u002Fsync behavior unless the phase explicitly\n  requires it.\n- Do not move the FEAT folder out of `03_IN_PROGRESS`.\n- Do not start dev servers, watch commands, or long-running UI processes.\n- Do not push to remotes.\n- Do not revert unrelated user changes.\n\n## Verification Rules\n\nApply the `serialized-build-commands` skill when available before running build,\ntest, lint, format, Cargo, pnpm, npm, or other shared-state commands.\n\nRun one verification command at a time. Wait for the result, inspect it, then\ndecide the next command.\n\nPrefer focused phase verification first. Run broad verification only when the\nphase or final checkpoint requires it and the narrower checks are green.\n\nDo not invent test results. Record exact commands and observed outcomes in the\nphase evidence.\n\nThe phase declarations are authoritative: needCodeReview and needTestCoverage are independent booleans; all four combinations are valid. Ordered native tasks and explicit configured checks retain their own obligations. Phase identifiers, positions, titles and file types never select gates. Developers can revise declarations with before\u002Fafter values, implemented scope, reason and evidence, without hiding failures. Reuse valid evidence for unchanged inputs. Ordinary missing tests or rejected reviews remain same-phase repair work; escalate only a concrete impasse or repeated lack of a meaningful repair path.\n\n## Phase Quality Gate Evidence\n\nAgents write evidence in phase documents. Hepha owns the machine-readable\nlifecycle fields (`**Status:**` and FeatureTasks Status) and Quality Gate\nDecision cells. Do not alter those reserved fields. New phase documents created\nby RefineFeature use this table:\n\n| Gate | Decision | Evidence \u002F Justification |\n| --- | --- | --- |\n| Changed files | missing | Exact production, test, and documentation paths changed in this phase. |\n| Tests | missing | Exact test files and commands, or a waiver explaining why tests are not useful for this phase. |\n| Gherkin\u002FPlaywright E2E | not applicable | Required only at the explicitly assigned workflow execution owner; retain EPIC E2E updates and execution links alongside phase TwinTests. |\n| Code review | missing | Phase review report path, or an explicit waiver and risk rationale. |\n\nRules:\n\n- Follow the declared coverage and review flags with scope reasons for N\u002FA. Execute independently required checks even when both flags are false.\n- UI changes may require frontend TwinTests and EPIC E2E updates. Run full-workflow E2E at its declared owner; a passing TwinTest never waives it.\n- Do not claim a phase is complete with a required `missing` gate. Supply the\n  missing evidence or justification and let Hepha retain the lifecycle state.\n- Do not mark a phase complete while an EPIC acceptance scenario assigned to\n  that phase lacks its planned executable evidence or an exact documented\n  existing-coverage mapping.\n\n## Quality-Gate Repair Mode\n\nWhen all implementation phase work is already completed but one or more phase\nquality gates are `missing`, do not redo completed implementation tasks. Work\nonly the missing gates:\n\n- Missing `Code review`: inspect the declared review obligation and existing report. When required, repair findings and obtain approval for the applicable scope. If scope changed, record a justified declaration revision; do not infer review applicability from file types or directly overwrite machine-owned decision cells.\n- Missing `Tests`: inspect the declared acceptance coverage and required commands. Reuse valid execution and assertion evidence, repair missing meaningful coverage, and execute missing or invalidated required checks. A justified declaration revision must preserve actual failures and the remaining obligations.\n- Missing `Gherkin\u002FPlaywright E2E`: inspect the assigned workflow obligation and original execution evidence. Repair and execute that obligation at its declared owner; isolated tests cannot replace a required full-workflow check.\n- Preserve completed phase implementation task checkboxes. Do not broaden scope\n  beyond the missing gate unless a review finding requires a fix.\n\n## Code Review Rules\n\nFor every code-relevant phase:\n\n1. Prefer a dedicated subagent Code Review Agent when a Pi subagent facility is\n   available, including `pi-subagents`, `@gotgenes\u002Fpi-subagents`, or an Agent\n   tool exposed by the active Pi profile.\n2. Give the reviewer the FEAT id, current phase, changed files from git\n   status\u002Fdiff, relevant phase docs, acceptance criteria, and exact verification\n   output. Do not ask the reviewer to re-run implementation.\n3. If no subagent facility is available, perform a separate review pass in the\n   current Pi session and record that fallback in the report.\n4. Write each review report under `\u003CFEAT folder>\u002Fcode-reviews\u002F` with a\n   phase-specific name such as `phase-2-code-review-\u003Ctimestamp>.md`.\n5. Treat BLOCKER, REQUIRED, and `NEEDS_CHANGES` findings as mandatory.\n6. For every finding, create or update a decision ledger recording fixed,\n   deferred, accepted-risk, rebutted, follow-up, or blocked with evidence.\n7. After fixing any blocking finding, rerun the smallest relevant verification\n   and rerun code review before advancing.\n8. Continue the fix-and-review loop until the latest report is APPROVED or has\n   no blocking findings, or until the same issue repeats after a documented\n   recovery attempt.\n\nDo not mark a phase complete while blocking review findings remain. Notes and\npolish items must be explicitly resolved as fixed, deferred, accepted-risk,\nrebutted, or follow-up.\n\n## Output Contract\n\nAt the end of the run, report:\n\n- project root and MemoryBank path used;\n- FEAT id and folder;\n- branch used;\n- whether the start transition was performed or already done;\n- phases completed\u002Fskipped during this run;\n- code-review reports created and rerun status;\n- verification commands and results;\n- blocker, if any;\n- changed files.\n\nKeep the final summary concise and concrete.\n","# Start Feature\n\n## Model Authority\n\nThis procedure is model-neutral. When invoked directly, execution is\n`direct_host` and remains in the current Pi, Codex, or Claude Code session; that\nactive host owns model selection. Do not query Hepha routing policy, request a\nmodel switch, automatically hand off, choose a fallback model, or fabricate\nroute-policy evidence. Direct execution does not fabricate an orchestrated\nreceipt. Only an explicit Hepha launcher or dashboard dispatch creates a\nseparate `orchestrated` worker, whose route is injected outside this skill.\n\nYou are executing the HEPHA `start-implementing` workflow directly from Pi.\n\nThis skill is for console-driven implementation startup and for HEPHA WebApp\nworkers that mount the same skill. It must work when Pi is opened from a\nregistered project or parent workspace and the user names the active project\nplus a FEAT id such as `FEAT-003`.\n\n## Objective\n\nResolve the active HEPHA project and MemoryBank, locate the requested FEAT,\nperform the start-feature transition when the FEAT is still READY, then in\nautonomous mode continue contract-declared phases with verification, code\nreview, required fixes, and review reruns until every declared phase is\nresolved or a real blocker requires human input.\n\n## Required Inputs\n\nAccept these from the user's message:\n\n- project name or alias, such as `HEPHA`, `Hepha`, or\n  `AgenticDevelopmentProcess`;\n- FEAT id, such as `FEAT-003`;\n- optional MemoryBank path override;\n- optional mode, where `autonomous` means continue through subsequent phases\n  without stopping after the start transition.\n\nIf the FEAT id is missing, ask one concise question. Otherwise, infer safely.\n\n## Workspace And Project Resolution\n\n1. Treat the current directory as the workspace starting point.\n2. Resolve the active project named by the user and locate its repository root.\n3. Read the project `AGENTS.md` and project brief or MemoryBank overview when\n   present before editing.\n4. Resolve the active project:\n   - `HEPHA`, `Hepha`, and `AgenticDevelopmentProcess` map to\n     `\u003Cworkspace root>\u002FAgenticDevelopmentProcess`.\n   - If the user names another project, use `docs\u002Fprojects.md` or the nearest\n     child repo folder matching that name.\n5. Read the child project `AGENTS.md` and `README.md` when present.\n6. Run project git, build, test, and edit commands from the child project root,\n   not from the parent workspace.\n\n## MemoryBank And FEAT Resolution\n\nResolve the MemoryBank in this order:\n\n1. User-supplied MemoryBank path, if present.\n2. `.hepha\u002Fprojects.json` in the child project, matching project `name`, `id`,\n   or child root path.\n3. `\u003Cchild project root>\u002FMemoryBank` when it contains `Features\u002F`.\n4. A direct search under the child root for `MemoryBank\u002FFeatures`.\n\nThen resolve the FEAT:\n\n1. Normalize the requested FEAT id to uppercase, for example `FEAT-003`.\n2. Search these folders under `\u003CMemoryBank>\u002FFeatures`:\n   - `02_READY_TO_DEVELOP` first;\n   - `03_IN_PROGRESS` second, for HEPHA WebApp runs that already performed the\n     folder move or direct runs that are being resumed;\n   - then `01_SUBMITTED`, `04_COMPLETED`, and `05_CANCELLED` only for\n     diagnostics.\n3. Match a FEAT folder by folder name or by `FeatureDescription.md` content.\n4. Continue only when the FEAT is in `02_READY_TO_DEVELOP` or\n   `03_IN_PROGRESS`.\n5. Require `FeatureDescription.md`, `FeatureTasks.md`,\n   `PhaseExecutionContract.json`, and every contract-declared phase document.\n6. Stop if the FEAT contains `[NEEDS VALIDATION]`, and report exact file and\n   line references.\n\n## Start Transition\n\nWhen the FEAT is in `02_READY_TO_DEVELOP`:\n\n1. Inspect `git status --short --branch` from the project root.\n2. Stop before changing branches if unrelated dirty files make the transition\n   unsafe.\n3. Create or switch to branch `feat\u002F\u003CFEAT-folder-name>` unless HEPHA-provided\n   runtime context names a different target branch.\n4. Move the FEAT folder to `Features\u002F03_IN_PROGRESS`, preserving the folder\n   name and files.\n5. Update `FeatureDescription.md` and `FeatureTasks.md` so status and folder\n   references say the FEAT is in progress.\n6. Update linked EPIC references when the parent EPIC is available.\n7. Create or update `start-feature-report.md` in the FEAT folder with branch,\n   source folder, target folder, timestamp, and validation findings.\n\nWhen the FEAT is already in `03_IN_PROGRESS`, do not move it back to READY.\nRepair stale READY\u002FSUBMITTED references in FEAT documents, ensure a\n`start-feature-report.md` exists, then continue from the first unresolved\ncontract phase.\n\nCreate focused checkpoint commits at meaningful gates when safe, especially for\nthe start transition and review-approved phase gates. Never push during this\nskill; `complete-feature` handles pushes, merges, and completed-folder moves\nafter human gates are accepted.\n\n## Start-Feature Post-Operation: Estimates\n\nBefore any implementation phase begins, calculate and write the following for\nevery contract-declared phase:\n\n- `Estimated Human Time`: focused engineering time for a competent developer;\n- `Estimated AI Time`: active agent\u002Fruntime time, excluding waiting and human\n  checkpoints.\n\nUse only parseable compact values: `30m`, `1h`, or a same-unit range such as\n`2-3h`. In ranges, use the literal ASCII hyphen-minus (`-`, U+002D) only; never\nuse an en dash (`–`, U+2013), em dash (`—`, U+2014), or another typographic dash.\nBefore returning, inspect every phase estimate and the `## Implementation Timing\nSummary`; replace any typographic range dash with the ASCII hyphen-minus. Add\n`## Implementation Timing Summary` to `FeatureTasks.md`, with full-FEAT totals\nfor both estimates. Do not invent actual duration: HEPHA records it from completed\nphase worker start\u002Fend timestamps. Stop and report a blocker if any contract-declared\nphase lacks both parseable estimates or the timing summary before implementation starts.\n\n## Phase Resolution\n\nBuild the phase inventory before editing. `PhaseExecutionContract.json` is the\nonly source of phase topology and ordering; never infer it from a phase number,\ntitle, or filename:\n\n1. Read `FeatureTasks.md`.\n2. Read and validate `PhaseExecutionContract.json`, then read exactly its\n   declared documents.\n3. Parse phase status from the `FeatureTasks.md` contract inventory first, then from\n   a phase file status line such as `**Status:** COMPLETED`.\n4. Normalize statuses:\n   - `Completed`, `Complete` -> `COMPLETED`\n   - `Skipped`, `N\u002FA` -> `SKIPPED`\n   - `Not Started`, `NOT_STARTED`, `Pending` -> `PENDING`\n   - `In Progress` -> `IN_PROGRESS`\n   - `Checkpoint In Progress` -> `CHECKPOINT_IN_PROGRESS`\n   - `Code Review In Progress` -> `CODE_REVIEW_IN_PROGRESS`\n   - `Awaiting Review` -> `AWAITING_REVIEW`\n   - `Awaiting User Acceptance` -> `AWAITING_USER_ACCEPTANCE`\n   - `Blocked`, `Failed`, `Rejected` -> `BLOCKED`\n5. Stop with a blocker if the execution contract is missing, invalid, or its\n   declared documents do not match the `Phases\u002F` directory.\n6. The current phase is the first contract-ordered phase that is not `COMPLETED` or\n   `SKIPPED`, unless a review-finding report clearly requires returning to a\n   specific earlier phase.\n7. If every contract phase is `COMPLETED` or `SKIPPED`, scan each phase's\n   `## Quality Gate Evidence` table. If any gate decision is `missing`, treat\n   the first contract-ordered phase with a missing gate as the current phase for\n   quality-gate repair only.\n\n## Autonomous Phase Loop\n\nWhen the user says `autonomous`, continue through phases in execution-contract order:\n\n1. Start at the current unresolved phase.\n2. Complete or skip only that phase's scoped work.\n3. Update durable task evidence and narrative notes in the phase document. Hepha owns lifecycle\u002Fstatus fields and FeatureTasks status cells.\n4. Run focused verification.\n5. Run code review when needCodeReview is true.\n6. Fix blocking review findings, rerun focused verification, and rerun code\n   review.\n7. Record completion or skip evidence only after verification and review gates\n   are clear; Hepha deterministically writes the lifecycle state.\n8. Rescan phases from disk.\n9. Continue to the next unresolved contract phase.\n\nStop only when:\n\n- all declared contract phases are `COMPLETED` or `SKIPPED`;\n- all phase quality gates are `satisfied`, `waived`, or `not applicable`;\n- a real blocker requires human judgment;\n- credentials, permissions, missing dependency, or unsafe action prevents safe\n  progress;\n- the same failure repeats after a documented recovery attempt;\n- the user requested interactive mode and the start transition is complete.\n\n## Per-Phase Work Rules\n\nBefore changing code in a phase:\n\n- Read the phase document.\n- Read `planning-analysis-report.md` whenever the current contract phase's\n  planning-index row identifies it as required context.\n- Read relevant parent EPIC and FEAT acceptance criteria, including linked\n  `EpicAcceptanceTests.md` scenarios and the current phase's `EPIC Acceptance\n  Traceability` assignments in `FeatureTasks.md` when present.\n- Read applicable `MemoryBank\u002FLessonsLearned` documents.\n- Inspect current git status from the child project root.\n- Preserve existing checked Markdown checklist items.\n- For a V3 phase, `PhaseExecutionContract.json` is the machine task authority.\n  The `## Phase Task Ledger` must be its exact checkbox projection: one entry\n  per contract task, same order, matching `[contract:\u003Cid>]` and\n  `[executor:\u003Cexecutor>]` markers. Never add an uncontracted checkbox there.\n  Put descriptive work under `## Detailed Work` as plain bullets.\n- If the phase has no checkbox ledger, add a compact `## Phase Task Ledger`\n  from its declared contract tasks before substantive work.\n\nDuring implementation:\n\n- Work only the current phase scope.\n- Skip checked tasks unless a changed file, failed verification, or review\n  finding explicitly invalidates them.\n- Do not edit reserved machine fields: `**Status:**`, FeatureTasks Status cells, or Quality Gate Evidence Decision cells. Hepha writes them from durable workflow events.\n- Record task progress and evidence; Hepha applies the lifecycle transition.\n- Do not broaden scope to dashboard\u002FUI\u002Fsync behavior unless the phase explicitly\n  requires it.\n- Do not move the FEAT folder out of `03_IN_PROGRESS`.\n- Do not start dev servers, watch commands, or long-running UI processes.\n- Do not push to remotes.\n- Do not revert unrelated user changes.\n\n## Verification Rules\n\nApply the `serialized-build-commands` skill when available before running build,\ntest, lint, format, Cargo, pnpm, npm, or other shared-state commands.\n\nRun one verification command at a time. Wait for the result, inspect it, then\ndecide the next command.\n\nPrefer focused phase verification first. Run broad verification only when the\nphase or final checkpoint requires it and the narrower checks are green.\n\nDo not invent test results. Record exact commands and observed outcomes in the\nphase evidence.\n\nThe phase declarations are authoritative: needCodeReview and needTestCoverage are independent booleans; all four combinations are valid. Ordered native tasks and explicit configured checks retain their own obligations. Phase identifiers, positions, titles and file types never select gates. Developers can revise declarations with before\u002Fafter values, implemented scope, reason and evidence, without hiding failures. Reuse valid evidence for unchanged inputs. Ordinary missing tests or rejected reviews remain same-phase repair work; escalate only a concrete impasse or repeated lack of a meaningful repair path.\n\n## Phase Quality Gate Evidence\n\nAgents write evidence in phase documents. Hepha owns the machine-readable\nlifecycle fields (`**Status:**` and FeatureTasks Status) and Quality Gate\nDecision cells. Do not alter those reserved fields. New phase documents created\nby RefineFeature use this table:\n\n| Gate | Decision | Evidence \u002F Justification |\n| --- | --- | --- |\n| Changed files | missing | Exact production, test, and documentation paths changed in this phase. |\n| Tests | missing | Exact test files and commands, or a waiver explaining why tests are not useful for this phase. |\n| Gherkin\u002FPlaywright E2E | not applicable | Required only at the explicitly assigned workflow execution owner; retain EPIC E2E updates and execution links alongside phase TwinTests. |\n| Code review | missing | Phase review report path, or an explicit waiver and risk rationale. |\n\nRules:\n\n- Follow the declared coverage and review flags with scope reasons for N\u002FA. Execute independently required checks even when both flags are false.\n- UI changes may require frontend TwinTests and EPIC E2E updates. Run full-workflow E2E at its declared owner; a passing TwinTest never waives it.\n- Do not claim a phase is complete with a required `missing` gate. Supply the\n  missing evidence or justification and let Hepha retain the lifecycle state.\n- Do not mark a phase complete while an EPIC acceptance scenario assigned to\n  that phase lacks its planned executable evidence or an exact documented\n  existing-coverage mapping.\n\n## Quality-Gate Repair Mode\n\nWhen all implementation phase work is already completed but one or more phase\nquality gates are `missing`, do not redo completed implementation tasks. Work\nonly the missing gates:\n\n- Missing `Code review`: inspect the declared review obligation and existing report. When required, repair findings and obtain approval for the applicable scope. If scope changed, record a justified declaration revision; do not infer review applicability from file types or directly overwrite machine-owned decision cells.\n- Missing `Tests`: inspect the declared acceptance coverage and required commands. Reuse valid execution and assertion evidence, repair missing meaningful coverage, and execute missing or invalidated required checks. A justified declaration revision must preserve actual failures and the remaining obligations.\n- Missing `Gherkin\u002FPlaywright E2E`: inspect the assigned workflow obligation and original execution evidence. Repair and execute that obligation at its declared owner; isolated tests cannot replace a required full-workflow check.\n- Preserve completed phase implementation task checkboxes. Do not broaden scope\n  beyond the missing gate unless a review finding requires a fix.\n\n## Code Review Rules\n\nFor every code-relevant phase:\n\n1. Prefer a dedicated subagent Code Review Agent when a Pi subagent facility is\n   available, including `pi-subagents`, `@gotgenes\u002Fpi-subagents`, or an Agent\n   tool exposed by the active Pi profile.\n2. Give the reviewer the FEAT id, current phase, changed files from git\n   status\u002Fdiff, relevant phase docs, acceptance criteria, and exact verification\n   output. Do not ask the reviewer to re-run implementation.\n3. If no subagent facility is available, perform a separate review pass in the\n   current Pi session and record that fallback in the report.\n4. Write each review report under `\u003CFEAT folder>\u002Fcode-reviews\u002F` with a\n   phase-specific name such as `phase-2-code-review-\u003Ctimestamp>.md`.\n5. Treat BLOCKER, REQUIRED, and `NEEDS_CHANGES` findings as mandatory.\n6. For every finding, create or update a decision ledger recording fixed,\n   deferred, accepted-risk, rebutted, follow-up, or blocked with evidence.\n7. After fixing any blocking finding, rerun the smallest relevant verification\n   and rerun code review before advancing.\n8. Continue the fix-and-review loop until the latest report is APPROVED or has\n   no blocking findings, or until the same issue repeats after a documented\n   recovery attempt.\n\nDo not mark a phase complete while blocking review findings remain. Notes and\npolish items must be explicitly resolved as fixed, deferred, accepted-risk,\nrebutted, or follow-up.\n\n## Output Contract\n\nAt the end of the run, report:\n\n- project root and MemoryBank path used;\n- FEAT id and folder;\n- branch used;\n- whether the start transition was performed or already done;\n- phases completed\u002Fskipped during this run;\n- code-review reports created and rerun status;\n- verification commands and results;\n- blocker, if any;\n- changed files.\n\nKeep the final summary concise and concrete.\n",{"name":5,"description":7,"agent_action":5},"master","57a527998225d4654bad4d3edb7380523849d28d","pi-packages\u002Fpi-skill-hepha-continue-implementation\u002Fskills\u002Fstart-feature","https:\u002F\u002Fgithub.com\u002Faboimpinto\u002FAgenticDevelopmentProcess","https:\u002F\u002Fgithub.com\u002Faboimpinto","2026-10-31 15:24:59",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":42},[32,35,38],{"tool":33,"command":34},"npx","npx skills add aboimpinto\u002FAgenticDevelopmentProcess --skill start-feature",{"tool":36,"command":37},"bunx","bunx skills add aboimpinto\u002FAgenticDevelopmentProcess --skill start-feature",{"tool":39,"command":40},"pnpm","pnpm dlx skills add aboimpinto\u002FAgenticDevelopmentProcess --skill start-feature","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Faboimpinto\u002FAgenticDevelopmentProcess\u002Fblob\u002Fmaster\u002Fpi-packages\u002Fpi-skill-hepha-continue-implementation\u002Fskills\u002Fstart-feature\u002FSKILL.md",true,[],[45,52,59,66,73,80],{"skillNo":46,"name":47,"nameZh":6,"description":48,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":49,"categories":50,"keywords":51,"publishedAt":19},"01M7C6RES661REA4K3SSD410VM","serialized-build-commands","Agents should invoke this skill before running build, test, lint, or format commands that use shared caches, file locks, build directories, or package manager state. Use especially before Rust Cargo commands such as cargo check, cargo test, cargo clippy, cargo fmt, cargo build, or any shell command containing cargo.","https:\u002F\u002Fgithub.com\u002Faboimpinto\u002FAgenticDevelopmentProcess\u002Fblob\u002Fmaster\u002Fpi-packages\u002Fpi-skill-serialized-build-commands\u002Fskills\u002Fserialized-build-commands\u002FSKILL.md",[],[],{"skillNo":53,"name":54,"nameZh":6,"description":55,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":56,"categories":57,"keywords":58,"publishedAt":19},"01M7C6R9EFN2C4620CFRMM5SMK","submit-epic","Use when the user asks Pi to submit, create, or draft a new HEPHA EPIC directly from a registered project workspace, preserving the useful legacy DevCycle MCP submit-epic command, for example \"submit-epic for HEPHA build observability projection hooks\" or \"create a HEPHA EPIC titled Runtime Approval Gates\".","https:\u002F\u002Fgithub.com\u002Faboimpinto\u002FAgenticDevelopmentProcess\u002Fblob\u002Fmaster\u002Fpi-packages\u002Fpi-skill-hepha-continue-implementation\u002Fskills\u002Fsubmit-epic\u002FSKILL.md",[],[],{"skillNo":60,"name":61,"nameZh":6,"description":62,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"publishedAt":19},"01M7C6R0NN4CAK6VPMPJH4Y5K3","refine-feature","Use when the user asks Pi to run refine-feature, refine a HEPHA FEAT, create FeatureTasks and phase files, or move a clarified FEAT to Ready To Develop, for example \"refine-feature FEAT-004\" or \"Use the refine-feature skill for HEPHA FEAT-004\".","https:\u002F\u002Fgithub.com\u002Faboimpinto\u002FAgenticDevelopmentProcess\u002Fblob\u002Fmaster\u002Fpi-packages\u002Fpi-skill-hepha-continue-implementation\u002Fskills\u002Frefine-feature\u002FSKILL.md",[],[],{"skillNo":67,"name":68,"nameZh":6,"description":69,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"publishedAt":19},"01M7C6QWEKWY4AEFDCBCXKE9E8","design-feature","Use when the user asks Pi to run design-feature, design a HEPHA FEAT, create UI requirements, or prepare UX\u002Fdesign artifacts for a FEAT, for example \"design-feature FEAT-004\" or \"Use the design-feature skill for HEPHA FEAT-004\".","https:\u002F\u002Fgithub.com\u002Faboimpinto\u002FAgenticDevelopmentProcess\u002Fblob\u002Fmaster\u002Fpi-packages\u002Fpi-skill-hepha-continue-implementation\u002Fskills\u002Fdesign-feature\u002FSKILL.md",[],[],{"skillNo":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":19},"01M7C6QSJEJ7GZP51CGKNPN27D","deep-dive","Use when the user asks Pi to run a HEPHA Deep-Dive for an EPIC or FEAT, gather Deep-Dive questions, or apply saved Deep-Dive answers to a MemoryBank document, for example \"use the deep-dive skill for HEPHA FEAT-004\".","https:\u002F\u002Fgithub.com\u002Faboimpinto\u002FAgenticDevelopmentProcess\u002Fblob\u002Fmaster\u002Fpi-packages\u002Fpi-skill-hepha-continue-implementation\u002Fskills\u002Fdeep-dive\u002FSKILL.md",[],[],{"skillNo":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":19},"01M7C6QMSSWNKS33Z4N2Z91H04","continue-implementation","Use when the user asks Pi to continue implementation, continue-implementing, resume implementation, or run a HEPHA FEAT implementation such as \"use the skill continue implementation for FEAT-002\" from a registered project workspace.","https:\u002F\u002Fgithub.com\u002Faboimpinto\u002FAgenticDevelopmentProcess\u002Fblob\u002Fmaster\u002Fpi-packages\u002Fpi-skill-hepha-continue-implementation\u002Fskills\u002Fcontinue-implementation\u002FSKILL.md",[],[]]