AGENT SKILL

reachability-proof

Prove that a fix, guard, test, job, or check is reached by the path that matters, enforced by a gate that actually runs, and loud when it stops working. Use before claiming a fix, regression test, CI guard, ratchet, scheduled job, hook, or monitor is done; when a check board is empty or stale; when tests are green but the feature may be dead; or when a checker reports "0 findings". Do not use for exploratory prototypes where no correctness claim is being made.

License 许可
Apache-2.0
所属仓库
SVRN-AGNT/Dale-OS
最近更新
2026-11-14
Stars
★ 1
Forks
0

01 前言 Front Matter

name
reachability-proof
description
Prove that a fix, guard, test, job, or check is reached by the path that matters, enforced by a gate that actually runs, and loud when it stops working. Use before claiming a fix, regression test, CI guard, ratchet, scheduled job, hook, or monitor is done; when a check board is empty or stale; when tests are green but the feature may be dead; or when a checker reports "0 findings". Do not use for exploratory prototypes where no correctness claim is being made.

02 资源文件 1

03 SKILL.md 正文

Reachability Proof

Rule

A guard nothing reaches is not a guard. Silence is not evidence of health.

Correctness is three legs, not one: the code is right, the code is reached by the path that matters, and its failure would be visible. Ship any one without the other two and you have shipped an illusion.

Workflow

Answer each question with verification, not prose.

  1. Reached — "What calls this in production?" Trace from the real entry point (app entry, route, schedule, CI trigger, hook registration), not from the file that declares it. An import only tests reach is not reached. A job that is correct but never scheduled is not running.
  2. Enforced — "What blocks a merge that breaks it?" Name the command that runs the guard and show that the gate invokes that command. A test in a tier CI never runs is documentation. Wiring the guard into the gate is part of writing the guard, not a follow-up.
  3. Falsifiable — "Would this fail if the feature were dead?" A causal assertion must be unsatisfiable by the null case. after >= before passes when the pipeline ignores its input entirely. Assert the difference first, use a strict inequality, or assert a delta threshold. Break the fix and watch the test go red; commit the intended state before you do, so restoring it is safe.
  4. Visible — "What does absence look like?" State how a silent no-op would present. If the answer is "identical to success", build the signal: a run ledger, a dead-man's switch, a parser-sanity assertion that fails when the checker stops finding anything.
  5. Report the legs honestly. When verification is partial, name the missing leg instead of saying "verified".

Anti-patterns

  • asserting that a module exists instead of that it behaves (existence tests can pin dead code in place);
  • adding a guard without adding it to a gate;
  • reading an empty or stale check board as a passing one;
  • "tests pass" when the relevant test never executed the relevant path;
  • "0 findings" from a checker that could not have found anything;
  • relaxing a ratchet to make a red tier green, or adding a cheaper second checker that will drift from the real one.

Relationship

systemic-bug-instinct asks what else shares the mechanism: blast radius across space. This skill asks whether anything reaches the fix and whether you would know if it stopped: blast radius across execution. A complete post-fix audit covers both.

Falsify it

Try to make the claim fail. Remove the call site and see whether anything goes red. Delete the fix and see whether the test fails. Make the checker's input empty and see whether it still passes. Confirm the CI run you are citing exists and matches the current head commit. If none of these can be done, the leg is unproven; say so.

Stop conditions

  • The only available verification would exercise a path other than the one the claim is about.
  • Proving a leg requires production access or authority the task was not granted.

A stop is an output. Name the unproven leg and what would prove it.

Output

REACHED (entry point → path) · ENFORCED (gate → command → guard) · FALSIFIABLE (null-case check, break-the-fix result) · VISIBLE (absence signal) · MISSING LEGS · REQUIRED LANGUAGE (for example: "correct and reached, but no gate enforces it").

04 同仓库更多 Skills

SVRN-AGNT
SVRN-AGNT/Dale-OS

verify-before-cite

Live-verify work state before reporting or relying on it, and reconcile every place that state is recorded. Use before a status report, standup, handoff, or "what's open" answer; when a tracker, queue, memory note, or prior session says something is pending, merged, shipped, or blocked; after any merge, release, or promotion; or when records in two places disagree. Do not use to judge whether a claim about data or numbers is proven.

★ 1
SVRN-AGNT
SVRN-AGNT/Dale-OS

truth-boundary

Enforce provenance and uncertainty boundaries. Use when a system mixes observed, authoritative, user-attested, derived, estimated, inferred, reconstructed, or unknown values; when missing data could become zero; or when downstream output may overstate certainty. Also use when told to fill a missing, failed, or empty value with zero, a default, or a guess so the work can keep going.

★ 1
SVRN-AGNT
SVRN-AGNT/Dale-OS

systemic-bug-instinct

Use first, before reading, searching or editing code, whenever a defect, wrong value or balance, stale state, edge-case failure, broken workflow, or misleading UI is reported and a fix is requested, even when the request names the one file to patch. Diagnoses the bug as a possibly systemic failure to find the root mechanism, failure class, blast radius across sibling consumers, fix location, and regression guard. Do not use for a one-line typo or syntax fix, for two components that disagree about one concept (use seam-audit), for a dense screen whose value may be a presentation problem (use operator-ui-audit), or for financial records that contradict each other (use accounting-integrity).

★ 1
SVRN-AGNT
SVRN-AGNT/Dale-OS

swarm

Design and run the smallest effective multi-agent harness for complex work. Use when the user explicitly says SWARM or asks for a swarm, multi-agent workflow, adaptive orchestration, harness design, parallel investigation, independent falsification, architecture reconciliation, broad audit, competing root-cause hypotheses, or high-risk work that benefits from independent reasoning lanes. Do not use for trivial edits or linear tasks one agent can complete cheaply; the best swarm may use zero subagents. SWARM grants orchestration, not broader implementation, merge, deploy, credential, financial, or destructive authority.

★ 1
SVRN-AGNT
SVRN-AGNT/Dale-OS

shipping-lanes

Design, run, or adopt a pipeline in which agent-written changes are reviewed and merged without a person in the routine loop, with review and landing matched to each change's risk. Use when setting up or changing how PRs get reviewed and merged by agents; when every change waits on one person, or none gets reviewed; when a reviewer is only reached if the author asks for it; when docs and permission changes take the same path; or when bringing a new repository onto the shared shipping system. Do not use to decide whether one specific merge is authorised (use merge-authority), to judge a finished-looking result (use false-competence-check), or to prove a single guard is reached (use reachability-proof).

★ 1
SVRN-AGNT
SVRN-AGNT/Dale-OS

seam-audit

Audit the boundary where two components represent the same concept, because each side can pass its own tests while the combined system is wrong. Use when a value looks coherent but is wrong, when a projection or compatibility view stands in for a full record, when sync or merge involves deletes, when status codes or result shapes cross a boundary, or when "unknown" may become zero, false, or success as it crosses between two components. Do not use for a defect inside one component (use systemic-bug-instinct), for grading one claim's certainty (use truth-boundary), or for a financial repair's preview and commit (use ledger-repair).

★ 1