[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7BYMFH2QFPFPJE4EC1500BK":3,"skill-same:01M7BYMFH2QFPFPJE4EC1500BK":43},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":23,"commitSha":24,"skillDirectory":25,"forksCount":13,"repositoryHtmlUrl":26,"repositoryHomepageUrl":6,"ownerHtmlUrl":27,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":28,"install":29,"resources":42},"01M7BYMFH2QFPFPJE4EC1500BK","adding-eval-scorecard",null,"Adopt the per-agent eval scorecard for a GAIA hub agent: write the harness→payload adapter, run the eval to produce a REAL scorecard, link + surface it from the agent's README, wire the release gate, and (for a new agent) generalize the format. Use when asked to 'add a scorecard', 'adopt the eval scorecard', 'generate the scorecard for \u003Cagent>', or wire scorecard CI for an agent. Builds on docs\u002Freference\u002Feval-scorecard.mdx and the email agent reference adapter.","secure-workflows\u002Fgaia","secure-workflows","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F311422533?v=4","01M7BYM925H8FAXM1DRD7F5DQD","01M7BYM92V01EZBAV87C93HNKX",0,"MIT","https:\u002F\u002Fgithub.com\u002Fsecure-workflows\u002Fgaia\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fadding-eval-scorecard\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: \"adding-eval-scorecard\"\ndescription: \"Adopt the per-agent eval scorecard for a GAIA hub agent: write the harness→payload adapter, run the eval to produce a REAL scorecard, link + surface it from the agent's README, wire the release gate, and (for a new agent) generalize the format. Use when asked to 'add a scorecard', 'adopt the eval scorecard', 'generate the scorecard for \u003Cagent>', or wire scorecard CI for an agent. Builds on docs\u002Freference\u002Feval-scorecard.mdx and the email agent reference adapter.\"\n---\n\n# Adding an Eval Scorecard to a GAIA Agent\n\nAdopt the release **eval scorecard** ([`docs\u002Freference\u002Feval-scorecard.mdx`](..\u002F..\u002F..\u002Fdocs\u002Freference\u002Feval-scorecard.mdx)) for one hub agent. The system is `harness → result payload → generator → scorecard`, with a standalone presence+regression release gate. The **email agent is the reference implementation** — mirror it.\n\n**Core modules (do not modify; reuse):**\n- `src\u002Fgaia\u002Feval\u002Frelease_scorecard.py` — `ResultPayload`, `compute_aggregate`, `render_scorecard`, `write_scorecard`, `validate_scorecard`, `carry_forward`. Harness-agnostic (stdlib + PyYAML only).\n- `src\u002Fgaia\u002Feval\u002Fscorecard_gate.py` — the standalone gate (`python -m gaia.eval.scorecard_gate`).\n- Reference adapter: `hub\u002Fagents\u002Femail\u002Fpython\u002Fpackaging\u002Fgen_scorecard.py`.\n\nThis is a **phased checklist with a hard gate at the real-eval step** — the scorecard MUST come from an actual eval run, never hand-authored numbers.\n\n## Phase 1 — Locate the agent's surfaces\n\n1. **Version source of truth** = the `version:` field in `\u003Cagent>\u002Fgaia-agent.yaml`. Never invent a parallel scheme.\n2. **Canonical README** (where the scorecard is linked + surfaced): for an npm-published agent it is the npm client README (e.g. `hub\u002Fagents\u002F\u003Cid>\u002Fnpm\u002FREADME.md`), NOT a `packaging\u002FREADME.md`. For a Python-only agent it is `hub\u002Fagents\u002F\u003Cid>\u002Fpython\u002FREADME.md`. Confirm which by checking what `release_agent_\u003Cid>.yml` publishes (`README:` env) — the published README is the one to link.\n3. **doc-root** = the directory holding that canonical README. The scorecard lives at `\u003Cdoc-root>\u002FSCORECARD.md` — a **single file updated in place**, versioned via the publish snapshot (same as README.md). **There is no `scorecards\u002F` directory.**\n4. **Eval vehicle**: what existing harness produces this agent's accuracy metric? (email → `gaia eval benchmark` over `tests\u002Ffixtures\u002Femail\u002F`.) If none exists, STOP and surface that — propose the minimal harness before building; do not invent numbers.\n\n   **Expect none.** Email is the only agent with a corpus + harness; for every other agent this step is the long pole, not a lookup. Two traps: an `agent_type` in the scenario corpus may name a **ChatAgent prompt profile** rather than your package (so a green scenario is measuring something else), and the ground truth itself is human judgment the factory does not automate. Build the corpus + deterministic fixture harness first — see [`porting-agent-to-hub`](..\u002Fporting-agent-to-hub\u002FSKILL.md) Phase 4 — then return here for the adapter and gate.\n\n## Phase 2 — Write the adapter (harness → payload)\n\nCopy `hub\u002Fagents\u002Femail\u002Fpython\u002Fpackaging\u002Fgen_scorecard.py` as the template. The adapter:\n- imports ONLY `gaia.eval.release_scorecard` (never the harness or agent package — preserve loose coupling);\n- reads the harness output, builds a `ResultPayload`;\n- populates `reproduction_command` with the **exact shell commands** to reproduce this scorecard, including all required env vars (`PYTHON_KEYRING_BACKEND`, `GAIA_AGENT_TOOL_TIMEOUT`, `PYTHONPATH`);\n- defines **\"judged\"** explicitly and **raises loudly** if zero results are judged (no silent 0.0);\n- records **dataset size** (total labeled examples) and **test_cases_run** (subset executed) as DISTINCT fields;\n- stores **repo-relative** paths only (never a local absolute path — it ships in a published artifact);\n- records the eval `limit`\u002Fconfig so future regression checks are comparable;\n- optionally populates `environment` (`gaia_commit`, `lemonade_version`, `model`, `hardware` — a class descriptor, never a hostname) and `breakdown` (`per_category` accuracy + `top_confusions`) — additive blocks that pin the run and explain the misses without ever affecting `aggregate.value`. Capture live environment (git SHA, server health version) in `main()`, not `build_payload`, so the payload builder stays pure and unit-testable;\n- writes to `\u003Cdoc-root>\u002FSCORECARD.md` (the single file; `--output-dir` overrides to a directory, but the filename is always `SCORECARD.md`).\n\nAdd an offline unit test against a committed sample harness-output fixture (see `tests\u002Ffixtures\u002Feval\u002Femail_benchmark_scorecard.json` + `tests\u002Funit\u002Feval\u002Ftest_release_scorecard.py::TestEmailAdapter`) so the adapter is testable without a live model.\n\n## Phase 3 — Run the REAL eval (hard gate — no hand-authored numbers)\n\nThe accuracy number must come from an actual run. For the email agent:\n\n```bash\n# Real eval needs Lemonade + the model. Prefer AMD hardware (Strix Halo \u002F Ryzen AI);\n# the [self-hosted, lemonade-eval] runner is the canonical environment.\nGAIA_AGENT_TOOL_TIMEOUT=1800 \\\nPYTHON_KEYRING_BACKEND=keyring.backends.null.Keyring \\\nPYTHONPATH=\"$(pwd)\" \\\n  \u003Cvenv>\u002Fbin\u002Fgaia eval benchmark \\\n    --model Gemma-4-E4B-it-GGUF \\\n    --mbox-path tests\u002Ffixtures\u002Femail\u002F_stub_inbox.mbox \\\n    --ground-truth tests\u002Ffixtures\u002Femail\u002F\u003Ctask>_ground_truth.json \\\n    --limit 220 --output-dir \u003Cpersistent-dir>\n\nPYTHONPATH=\"$(pwd)\" \\\n\u003Cvenv>\u002Fbin\u002Fpython hub\u002Fagents\u002Femail\u002Fpython\u002Fpackaging\u002Fgen_scorecard.py \\\n    --benchmark-dir \u003Cpersistent-dir> --limit 220\n# → writes hub\u002Fagents\u002Femail\u002Fnpm\u002FSCORECARD.md in place\n```\n\n⚠️ **Run evals SERIALLY — never two at once** (CLAUDE.md). A scorecard refresh loop is the\nmost likely thing to violate this: chain runs with `&&`, never background them with `&`.\nConcurrent runs race-evict each other's models on the single-tenant Lemonade backend and\nproduce garbage (`ctx_size` errors, spurious `INFRA_ERROR`) that looks like a real\nregression. Before each run: `ps aux | grep \"gaia eval\" | grep -v grep | wc -l` must be 0.\n\n**Pass both fixture flags explicitly.** `ls tests\u002Ffixtures\u002Femail\u002F` and pick the real files\n— ground truth is **task-specific** (`action_items_`, `briefing_`, `drafting_`,\n`followups_`, `longthread_`), and the CLI's `--ground-truth` default points at a\n`ground_truth.json` that does not exist. Omit the flag and the run prints one `[WARN]` and\nscores nothing, which reads as a successful eval right up until the scorecard is empty.\n\n**Headless gotchas (see memory `project-email-benchmark-headless-gotchas`):**\n- `PYTHON_KEYRING_BACKEND=keyring.backends.null.Keyring` — the email agent's calendar-connector resolution blocks forever on the macOS Keychain (and can stall on Linux SecretService) in non-interactive contexts. Without this it hangs at 0% CPU during agent construction.\n- `PYTHONPATH=\"$(pwd)\"` — the benchmark imports `tests.fixtures.email.*`; the console script doesn't add the repo root.\n- `GAIA_AGENT_TOOL_TIMEOUT=1800` — triage of the full corpus is one tool call (~17 min for ~100 emails on a 4B model); a lower timeout (the 180s default, or even 900s) abandons it mid-run, yielding a degenerate 0-email FAIL run.\n- Write `--output-dir` to a **persistent** dir, not `\u002Ftmp` (cleared on session resume).\n- Record honestly: if the metric is low for a known reason (e.g. a taxonomy\u002Flabel mismatch), put the explanation in the adapter's `methodology` string and link the tracking issue — never inflate the number.\n\n## Phase 4 — Surface, link, and gate\n\n1. **Link + surface** from the canonical README: a one-line `Eval scorecard (vX.Y.Z): aggregate N\u002F100 … ([.\u002FSCORECARD.md](.\u002FSCORECARD.md))`. The relative link must resolve in-repo.\n2. **npm `files`**: if the agent publishes on npm, add `SCORECARD.md` to `package.json` `files`. **Do not** add a `scorecards\u002F` directory — only the single current file ships.\n3. **Hub display**: a published scorecard surfaces on the agent's hub page \u002F Agent UI detail view (see `workers\u002Fagent-hub` + `AgentDetailModal.tsx`); ensure the publish step passes `--eval-scorecard \u003Cdoc-root>\u002FSCORECARD.md` to `publish_to_r2.py`.\n4. **Release gate**: add a `scorecard-gate` job to `release_agent_\u003Cid>.yml` and list it in `publish.needs`. The job runs on a GitHub-hosted runner (it only parses committed files — no eval):\n   ```bash\n   # Presence-only (no previous tag yet):\n   python -m gaia.eval.scorecard_gate \\\n     --scorecard \u003Cdoc-root>\u002FSCORECARD.md\n\n   # With best-effort previous-release baseline (recommended for CI):\n   PREV=\"$(git describe --tags --abbrev=0 --match 'agent-pkg-\u003Cid>-*' \"${GITHUB_REF_NAME}^\" 2>\u002Fdev\u002Fnull || true)\"\n   if [ -n \"$PREV\" ]; then\n     python -m gaia.eval.scorecard_gate \\\n       --scorecard \u003Cdoc-root>\u002FSCORECARD.md --baseline-ref \"$PREV\"\n   else\n     python -m gaia.eval.scorecard_gate \\\n       --scorecard \u003Cdoc-root>\u002FSCORECARD.md\n   fi\n   ```\n   The job must NOT have `continue-on-error`, an `environment:`, or a `permissions:` override (inherits `contents: read`; needs no secrets). Fetch full history (`fetch-depth: 0`) so `git describe` resolves.\n5. **Auto-update\u002Freject loop**: for re-running on agent changes and refreshing the committed scorecard, see `eval-scorecard.mdx` \"Keeping the scorecard current\" and the self-hosted refresh workflow — reject-on-worse is the gate; better-or-equal refreshes the committed `SCORECARD.md`.\n\n## Phase 5 — Verify (evidence before \"done\")\n\nRun and capture: the generated `SCORECARD.md`; the gate **passing** on it (exit 0); the gate **blocking** a manufactured regression (exit 1, via `--baseline-file` with a higher-scoring card) and a missing card (exit 1); a by-hand recompute of the aggregate from `aggregate.components` matching the recorded value. Run `python util\u002Flint.py --all` and the eval unit tests. These are the PR's real-world proof.\n\n## Versioning\n\n- **Patch** release → `carry_forward(prev_scorecard_path, new_version)` reads the version from the front matter of the current `SCORECARD.md` (not from the filename) and copies results verbatim, sets `inherited_from`; do NOT re-run the eval.\n- **Minor\u002Fmajor** release → re-run the eval (Phase 3); `carry_forward` refuses a non-patch bump with a \"re-run\" error.\n\n## Output style\n\nFollow [CLAUDE.md](..\u002F..\u002F..\u002FCLAUDE.md) → \"How You Communicate\".\n","# Adding an Eval Scorecard to a GAIA Agent\n\nAdopt the release **eval scorecard** ([`docs\u002Freference\u002Feval-scorecard.mdx`](..\u002F..\u002F..\u002Fdocs\u002Freference\u002Feval-scorecard.mdx)) for one hub agent. The system is `harness → result payload → generator → scorecard`, with a standalone presence+regression release gate. The **email agent is the reference implementation** — mirror it.\n\n**Core modules (do not modify; reuse):**\n- `src\u002Fgaia\u002Feval\u002Frelease_scorecard.py` — `ResultPayload`, `compute_aggregate`, `render_scorecard`, `write_scorecard`, `validate_scorecard`, `carry_forward`. Harness-agnostic (stdlib + PyYAML only).\n- `src\u002Fgaia\u002Feval\u002Fscorecard_gate.py` — the standalone gate (`python -m gaia.eval.scorecard_gate`).\n- Reference adapter: `hub\u002Fagents\u002Femail\u002Fpython\u002Fpackaging\u002Fgen_scorecard.py`.\n\nThis is a **phased checklist with a hard gate at the real-eval step** — the scorecard MUST come from an actual eval run, never hand-authored numbers.\n\n## Phase 1 — Locate the agent's surfaces\n\n1. **Version source of truth** = the `version:` field in `\u003Cagent>\u002Fgaia-agent.yaml`. Never invent a parallel scheme.\n2. **Canonical README** (where the scorecard is linked + surfaced): for an npm-published agent it is the npm client README (e.g. `hub\u002Fagents\u002F\u003Cid>\u002Fnpm\u002FREADME.md`), NOT a `packaging\u002FREADME.md`. For a Python-only agent it is `hub\u002Fagents\u002F\u003Cid>\u002Fpython\u002FREADME.md`. Confirm which by checking what `release_agent_\u003Cid>.yml` publishes (`README:` env) — the published README is the one to link.\n3. **doc-root** = the directory holding that canonical README. The scorecard lives at `\u003Cdoc-root>\u002FSCORECARD.md` — a **single file updated in place**, versioned via the publish snapshot (same as README.md). **There is no `scorecards\u002F` directory.**\n4. **Eval vehicle**: what existing harness produces this agent's accuracy metric? (email → `gaia eval benchmark` over `tests\u002Ffixtures\u002Femail\u002F`.) If none exists, STOP and surface that — propose the minimal harness before building; do not invent numbers.\n\n   **Expect none.** Email is the only agent with a corpus + harness; for every other agent this step is the long pole, not a lookup. Two traps: an `agent_type` in the scenario corpus may name a **ChatAgent prompt profile** rather than your package (so a green scenario is measuring something else), and the ground truth itself is human judgment the factory does not automate. Build the corpus + deterministic fixture harness first — see [`porting-agent-to-hub`](..\u002Fporting-agent-to-hub\u002FSKILL.md) Phase 4 — then return here for the adapter and gate.\n\n## Phase 2 — Write the adapter (harness → payload)\n\nCopy `hub\u002Fagents\u002Femail\u002Fpython\u002Fpackaging\u002Fgen_scorecard.py` as the template. The adapter:\n- imports ONLY `gaia.eval.release_scorecard` (never the harness or agent package — preserve loose coupling);\n- reads the harness output, builds a `ResultPayload`;\n- populates `reproduction_command` with the **exact shell commands** to reproduce this scorecard, including all required env vars (`PYTHON_KEYRING_BACKEND`, `GAIA_AGENT_TOOL_TIMEOUT`, `PYTHONPATH`);\n- defines **\"judged\"** explicitly and **raises loudly** if zero results are judged (no silent 0.0);\n- records **dataset size** (total labeled examples) and **test_cases_run** (subset executed) as DISTINCT fields;\n- stores **repo-relative** paths only (never a local absolute path — it ships in a published artifact);\n- records the eval `limit`\u002Fconfig so future regression checks are comparable;\n- optionally populates `environment` (`gaia_commit`, `lemonade_version`, `model`, `hardware` — a class descriptor, never a hostname) and `breakdown` (`per_category` accuracy + `top_confusions`) — additive blocks that pin the run and explain the misses without ever affecting `aggregate.value`. Capture live environment (git SHA, server health version) in `main()`, not `build_payload`, so the payload builder stays pure and unit-testable;\n- writes to `\u003Cdoc-root>\u002FSCORECARD.md` (the single file; `--output-dir` overrides to a directory, but the filename is always `SCORECARD.md`).\n\nAdd an offline unit test against a committed sample harness-output fixture (see `tests\u002Ffixtures\u002Feval\u002Femail_benchmark_scorecard.json` + `tests\u002Funit\u002Feval\u002Ftest_release_scorecard.py::TestEmailAdapter`) so the adapter is testable without a live model.\n\n## Phase 3 — Run the REAL eval (hard gate — no hand-authored numbers)\n\nThe accuracy number must come from an actual run. For the email agent:\n\n```bash\n# Real eval needs Lemonade + the model. Prefer AMD hardware (Strix Halo \u002F Ryzen AI);\n# the [self-hosted, lemonade-eval] runner is the canonical environment.\nGAIA_AGENT_TOOL_TIMEOUT=1800 \\\nPYTHON_KEYRING_BACKEND=keyring.backends.null.Keyring \\\nPYTHONPATH=\"$(pwd)\" \\\n  \u003Cvenv>\u002Fbin\u002Fgaia eval benchmark \\\n    --model Gemma-4-E4B-it-GGUF \\\n    --mbox-path tests\u002Ffixtures\u002Femail\u002F_stub_inbox.mbox \\\n    --ground-truth tests\u002Ffixtures\u002Femail\u002F\u003Ctask>_ground_truth.json \\\n    --limit 220 --output-dir \u003Cpersistent-dir>\n\nPYTHONPATH=\"$(pwd)\" \\\n\u003Cvenv>\u002Fbin\u002Fpython hub\u002Fagents\u002Femail\u002Fpython\u002Fpackaging\u002Fgen_scorecard.py \\\n    --benchmark-dir \u003Cpersistent-dir> --limit 220\n# → writes hub\u002Fagents\u002Femail\u002Fnpm\u002FSCORECARD.md in place\n```\n\n⚠️ **Run evals SERIALLY — never two at once** (CLAUDE.md). A scorecard refresh loop is the\nmost likely thing to violate this: chain runs with `&&`, never background them with `&`.\nConcurrent runs race-evict each other's models on the single-tenant Lemonade backend and\nproduce garbage (`ctx_size` errors, spurious `INFRA_ERROR`) that looks like a real\nregression. Before each run: `ps aux | grep \"gaia eval\" | grep -v grep | wc -l` must be 0.\n\n**Pass both fixture flags explicitly.** `ls tests\u002Ffixtures\u002Femail\u002F` and pick the real files\n— ground truth is **task-specific** (`action_items_`, `briefing_`, `drafting_`,\n`followups_`, `longthread_`), and the CLI's `--ground-truth` default points at a\n`ground_truth.json` that does not exist. Omit the flag and the run prints one `[WARN]` and\nscores nothing, which reads as a successful eval right up until the scorecard is empty.\n\n**Headless gotchas (see memory `project-email-benchmark-headless-gotchas`):**\n- `PYTHON_KEYRING_BACKEND=keyring.backends.null.Keyring` — the email agent's calendar-connector resolution blocks forever on the macOS Keychain (and can stall on Linux SecretService) in non-interactive contexts. Without this it hangs at 0% CPU during agent construction.\n- `PYTHONPATH=\"$(pwd)\"` — the benchmark imports `tests.fixtures.email.*`; the console script doesn't add the repo root.\n- `GAIA_AGENT_TOOL_TIMEOUT=1800` — triage of the full corpus is one tool call (~17 min for ~100 emails on a 4B model); a lower timeout (the 180s default, or even 900s) abandons it mid-run, yielding a degenerate 0-email FAIL run.\n- Write `--output-dir` to a **persistent** dir, not `\u002Ftmp` (cleared on session resume).\n- Record honestly: if the metric is low for a known reason (e.g. a taxonomy\u002Flabel mismatch), put the explanation in the adapter's `methodology` string and link the tracking issue — never inflate the number.\n\n## Phase 4 — Surface, link, and gate\n\n1. **Link + surface** from the canonical README: a one-line `Eval scorecard (vX.Y.Z): aggregate N\u002F100 … ([.\u002FSCORECARD.md](.\u002FSCORECARD.md))`. The relative link must resolve in-repo.\n2. **npm `files`**: if the agent publishes on npm, add `SCORECARD.md` to `package.json` `files`. **Do not** add a `scorecards\u002F` directory — only the single current file ships.\n3. **Hub display**: a published scorecard surfaces on the agent's hub page \u002F Agent UI detail view (see `workers\u002Fagent-hub` + `AgentDetailModal.tsx`); ensure the publish step passes `--eval-scorecard \u003Cdoc-root>\u002FSCORECARD.md` to `publish_to_r2.py`.\n4. **Release gate**: add a `scorecard-gate` job to `release_agent_\u003Cid>.yml` and list it in `publish.needs`. The job runs on a GitHub-hosted runner (it only parses committed files — no eval):\n   ```bash\n   # Presence-only (no previous tag yet):\n   python -m gaia.eval.scorecard_gate \\\n     --scorecard \u003Cdoc-root>\u002FSCORECARD.md\n\n   # With best-effort previous-release baseline (recommended for CI):\n   PREV=\"$(git describe --tags --abbrev=0 --match 'agent-pkg-\u003Cid>-*' \"${GITHUB_REF_NAME}^\" 2>\u002Fdev\u002Fnull || true)\"\n   if [ -n \"$PREV\" ]; then\n     python -m gaia.eval.scorecard_gate \\\n       --scorecard \u003Cdoc-root>\u002FSCORECARD.md --baseline-ref \"$PREV\"\n   else\n     python -m gaia.eval.scorecard_gate \\\n       --scorecard \u003Cdoc-root>\u002FSCORECARD.md\n   fi\n   ```\n   The job must NOT have `continue-on-error`, an `environment:`, or a `permissions:` override (inherits `contents: read`; needs no secrets). Fetch full history (`fetch-depth: 0`) so `git describe` resolves.\n5. **Auto-update\u002Freject loop**: for re-running on agent changes and refreshing the committed scorecard, see `eval-scorecard.mdx` \"Keeping the scorecard current\" and the self-hosted refresh workflow — reject-on-worse is the gate; better-or-equal refreshes the committed `SCORECARD.md`.\n\n## Phase 5 — Verify (evidence before \"done\")\n\nRun and capture: the generated `SCORECARD.md`; the gate **passing** on it (exit 0); the gate **blocking** a manufactured regression (exit 1, via `--baseline-file` with a higher-scoring card) and a missing card (exit 1); a by-hand recompute of the aggregate from `aggregate.components` matching the recorded value. Run `python util\u002Flint.py --all` and the eval unit tests. These are the PR's real-world proof.\n\n## Versioning\n\n- **Patch** release → `carry_forward(prev_scorecard_path, new_version)` reads the version from the front matter of the current `SCORECARD.md` (not from the filename) and copies results verbatim, sets `inherited_from`; do NOT re-run the eval.\n- **Minor\u002Fmajor** release → re-run the eval (Phase 3); `carry_forward` refuses a non-patch bump with a \"re-run\" error.\n\n## Output style\n\nFollow [CLAUDE.md](..\u002F..\u002F..\u002FCLAUDE.md) → \"How You Communicate\".\n",{"name":5,"description":7},"main","11039b64e036dca33313350b4cd1d6c2a791eb15",".claude\u002Fskills\u002Fadding-eval-scorecard","https:\u002F\u002Fgithub.com\u002Fsecure-workflows\u002Fgaia","https:\u002F\u002Fgithub.com\u002Fsecure-workflows","2026-10-26 10:48:34",{"source":8,"skillName":5,"commands":30,"prompt":40,"zipAvailable":41},[31,34,37],{"tool":32,"command":33},"npx","npx skills add secure-workflows\u002Fgaia --skill adding-eval-scorecard",{"tool":35,"command":36},"bunx","bunx skills add secure-workflows\u002Fgaia --skill adding-eval-scorecard",{"tool":38,"command":39},"pnpm","pnpm dlx skills add secure-workflows\u002Fgaia --skill adding-eval-scorecard","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fsecure-workflows\u002Fgaia\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fadding-eval-scorecard\u002FSKILL.md",true,[],[44,51,58,65,72,79],{"skillNo":45,"name":46,"nameZh":6,"description":47,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":48,"categories":49,"keywords":50,"publishedAt":18},"01M7BYXGPPQJGVVD0DPS2WK3FS","web-search","Search the web via the Brave Search API. Use when the user asks about current events or external facts.","https:\u002F\u002Fgithub.com\u002Fsecure-workflows\u002Fgaia\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Fweb-search\u002FSKILL.md",[],[],{"skillNo":52,"name":53,"nameZh":6,"description":54,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":55,"categories":56,"keywords":57,"publishedAt":18},"01M7BYXDF2TM5JPMEQKMMHW0VJ","triage-support-ticket","Triage an inbound support ticket end to end. Use when the user pastes a support ticket or asks to triage one.","https:\u002F\u002Fgithub.com\u002Fsecure-workflows\u002Fgaia\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Ftriage-support-ticket\u002FSKILL.md",[],[],{"skillNo":59,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":18},"01M7BYXAHWNBMJ8Y1Q51EZ53K7","tool-mismatch","Declares a tool its tools.py never registers. Used to prove validation fails loudly with no partial load.","https:\u002F\u002Fgithub.com\u002Fsecure-workflows\u002Fgaia\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Ftool-mismatch\u002FSKILL.md",[],[],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":18},"01M7BYX6CZ63AM6HB90PPNXKF7","report-archive","Archive a generated report to the user's reports folder and register it with the team's report index. Use when the user asks to file, archive, or publish a finished report.","https:\u002F\u002Fgithub.com\u002Fsecure-workflows\u002Fgaia\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Freport-archive\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":18},"01M7BYX4PM4HGNZ0QBZ54BTG1N","local-capability","Writes files on the local workspace. Used to prove Phase 1 refuses local-capability permissions.","https:\u002F\u002Fgithub.com\u002Fsecure-workflows\u002Fgaia\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Flocal-capability\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":18},"01M7BYX1KVPKHHZ5Z3R3BB0QQA","incident-review","Run an incident postmortem. Stands in for a Claude Code skill library imported read-only from .claude\u002Fskills\u002F.","https:\u002F\u002Fgithub.com\u002Fsecure-workflows\u002Fgaia\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Fincident-review\u002FSKILL.md",[],[]]