← ClaudeAtlas

with-groundinglisted

Use as the minimal v8-conformant project-grounded fixture for skill-graph package tests. Activate this skill when verifying that lint enforces deployment_target: project requiring grounding fields (subject_matter, grounding_mode, truth_sources, failure_modes, evidence_priority). Do NOT use as a production skill (use a real grounded skill from the canonical library).
jacob-balslev/skill-graph · ★ 0 · AI & Automation · score 66
Install: claude install-skill jacob-balslev/skill-graph
# With-Grounding Fixture This fixture exercises the v8 rule `deployment_target: project → grounding required`. It also carries `keywords` so routing-quality checks have realistic activation metadata to inspect, but `keywords` are not schema-required. ## Coverage The five required `grounding` sub-fields (`subject_matter`, `grounding_mode`, `truth_sources`, `failure_modes`, `evidence_priority`), an object-shaped truth_source entry with `path` + `anchor` + `note`, and a non-empty `keywords` array for routing realism. No relations, no Audit Status, no Understanding fields. ## Philosophy The grounding block is the contract that distinguishes a knowledge skill from a hallucinated one. A `deployment_target: project` skill is making claims about specific local files; those claims rot when the files move. The grounding block records *which* files anchor the claim, *how* to detect drift (anchor or line_range hashing), and *what* failure modes appear when drift occurs. This fixture is the smallest configuration that exercises every grounding required field without depending on a sibling skills clone. ## Verification ```bash node scripts/skill-lint.js --path examples/fixture-skills/with-grounding # expected: 0 errors ``` Lint must accept this fixture even when the published `@skill-graph/cli` package is installed standalone (no sibling `../skills/skills/` clone). The truth_source path resolves under the package's own `examples/fixture-skills/` tree, not under the canonical skill