Source Grounding
Standing rule — active every response. No invocation needed for routine use.
Consent gates (G1–G4, declared)
| # |
what it gates |
when it fires |
Agent lane |
Hook lane |
| G1 |
model tier |
before work starts |
ask_question, 3 tiers, wait for pick |
suppressed — auto-Light, no tier question |
| G2 |
how to proceed on an unfetchable source |
source can't be fetched, user present |
ask_question |
ask_question |
| G3 |
entanglement hand-off |
after the findings, cross-domain |
ask_question, once |
ask_question, once |
| G4 |
self error-report |
skill misbehaves |
offer, never auto-submit |
offer, never auto-submit |
Hook cells assume an interactive session (G2–G4 need a user to answer); non-interactive Hook fires D1 instead of G2, and offers nothing for G3/G4.
What to verify (not memory)
- CRITICAL (always fetch or flag — P2): API/SDK call signatures · library versions & deprecations · CVEs/security advisories · auth/crypto specs · LLM model IDs & params
- MEDIUM (verify when unsure): package names · config keys · CLI flags · protocol specs
- LOW/stable: math, algorithms, language syntax → memory fine (P3)
How
- Identify the version-sensitive claim.
- Name the authoritative source (official docs, advisory DB, package registry, spec, source code).
- Fetch (WebSearch/WebFetch/docs MCP) — or flag
⚠️ unverified: check [source] (D1).
- Cite at CRITICAL/MEDIUM. Don't over-verify stable facts (P3).
Per-claim-type authoritative source map: read references/sources.md when choosing where to verify.
Source hierarchy (1 = strongest)
- Source code / spec / RFC
- Official/vendor docs — authoritative secondary (honor
.coalmine.json trustedDomains if set: treat those domains as additional authoritative / tier-2 sources)
- Multiple reputable third-party sources
- Single blog — corroborate first (P4)
- Training memory — weakest for volatile facts
Why each rank sits where it does: references/sources.md.
Non-interactive runs: log unfetchable claims as ⚠️ UNVERIFIED and continue (D1). Interactive: when sources cannot be fetched, confirm how to proceed via ask_question (G2).
Prohibitions (P1–P6, declared)
| # |
never … |
| P1 |
default to English just because this file is English |
| P2 |
skip fetching or flagging a CRITICAL version-sensitive claim |
| P3 |
over-verify a stable/LOW fact |
| P4 |
cite a single blog source without corroborating first |
| P5 |
auto-submit the self error-report |
| P6 |
include unapproved code or paths in the self error-report |
The shared footer's never fix without a chosen option does not apply here — this skill defines no Fix mode section, so that clause resolves vacuously; not counted above.
Degrade paths (D1–D4, declared)
| # |
branch |
fires when: |
| D1 |
log as ⚠️ UNVERIFIED, continue, never block |
non-interactive, source unfetchable |
| D2 |
degrade to model tier + reasoning depth, never fake parallelism |
no capability lever for the target tier on this host |
| D3 |
fall back to a numbered text menu |
host has no question tool |
| D4 |
fixed at Light, no tier question, no sub-agents |
Hook Context (auto-triggered) |
This ledger deliberately diverges from skill-authoring.md §3b's column-or-separate-ledger rule for lane-applicability — gold-standard keeps a lane column, but here a column asserting a lane value that contradicted its own row's condition text measured worse than no column at all (SKILL-VARIANCE-WALK.md §Run 43: a bimodal Hook-Q4 split, the pre-registered key matching neither camp). D2 is restated at four sites in the shared partials — the general clause, the Standard row's "(else single-agent)", the Heavy row's "if supported", and the Heavy-specific "escalate by model + reasoning only" — all one row. D3 is stated once, in the shared Escalation footer's question-tool list ("…none → numbered text menu"). Neither is a new branch. D4's own branch text is restated verbatim in the footer's Hook Context line — same row, not a new one. The Freshness cap (scope already audited this session → cap at Light) is a tier-selection modifier on G1, not a degrade branch. The footer's Fix-mode-dependent offer clause is not a fifth branch — this skill defines no Fix mode section, so it never fires.
Output — 2 locations, declared
A location is a place this skill writes something a reader can see; the absence of an annotation is not one.
- Verified:
✅ [claim] — source: [link/file]
- Unverified:
⚠️ unverified — check [exact source]
Stable fact: no annotation is written — not a location, not counted above.
AUTHORITATIVE vs DIVERSE
- AUTHORITATIVE (one ground truth): API/version/config/spec → go to the actual source code or official docs.
- DIVERSE (triangulate ≥ 3): "what's best" / landscape / patterns → multiple repos + docs + community; note conflicts.
1---2name: source-grounding3description: Verify version-sensitive facts against live authoritative sources before asserting them in code or answers. Triggers on: "/source-grounding", "source-grounding", "sourcing". Standing rule — always active via CLAUDE.md. Invoke for deep verification work (API signatures, CVEs, model IDs, auth flows, deprecated patterns, security advisories).4---56# Source Grounding78<!-- SHARED:LANGUAGE_HEADER -->910Standing rule — active every response. No invocation needed for routine use.1112## Consent gates (G1–G4, declared)13| # | what it gates | when it fires | Agent lane | Hook lane |14|---|---|---|---|---|15| G1 | model tier | before work starts | `ask_question`, 3 tiers, wait for pick | suppressed — auto-Light, no tier question |16| G2 | how to proceed on an unfetchable source | source can't be fetched, user present | `ask_question` | `ask_question` |17| G3 | entanglement hand-off | after the findings, cross-domain | `ask_question`, once | `ask_question`, once |18| G4 | self error-report | skill misbehaves | offer, never auto-submit | offer, never auto-submit |1920Hook cells assume an interactive session (G2–G4 need a user to answer); non-interactive Hook fires D1 instead of G2, and offers nothing for G3/G4.2122## What to verify (not memory)23- **CRITICAL** (always fetch or flag — P2): API/SDK call signatures · library versions & deprecations · CVEs/security advisories · auth/crypto specs · LLM model IDs & params24- **MEDIUM** (verify when unsure): package names · config keys · CLI flags · protocol specs25- **LOW/stable**: math, algorithms, language syntax → memory fine (P3)2627## How281. Identify the version-sensitive claim.292. Name the authoritative source (official docs, advisory DB, package registry, spec, source code).303. Fetch (WebSearch/WebFetch/docs MCP) — or flag `⚠️ unverified: check [source]` (D1).314. Cite at CRITICAL/MEDIUM. Don't over-verify stable facts (P3).3233Per-claim-type authoritative source map: read `references/sources.md` when choosing where to verify.3435## Source hierarchy (1 = strongest)361. Source code / spec / RFC372. Official/vendor docs — authoritative secondary (honor `.coalmine.json` `trustedDomains` if set: treat those domains as additional authoritative / tier-2 sources)383. Multiple reputable third-party sources394. Single blog — corroborate first (P4)405. Training memory — weakest for volatile facts4142Why each rank sits where it does: `references/sources.md`.4344Non-interactive runs: log unfetchable claims as `⚠️ UNVERIFIED` and continue (D1). Interactive: when sources cannot be fetched, confirm how to proceed via `ask_question` (G2).4546## Prohibitions (P1–P6, declared)47| # | never … |48|---|---|49| P1 | default to English just because this file is English |50| P2 | skip fetching or flagging a CRITICAL version-sensitive claim |51| P3 | over-verify a stable/LOW fact |52| P4 | cite a single blog source without corroborating first |53| P5 | auto-submit the self error-report |54| P6 | include unapproved code or paths in the self error-report |5556The shared footer's `never fix without a chosen option` does not apply here — this skill defines no Fix mode section, so that clause resolves vacuously; not counted above.5758## Degrade paths (D1–D4, declared)59| # | branch | fires when: |60|---|---|---|61| D1 | log as `⚠️ UNVERIFIED`, continue, never block | non-interactive, source unfetchable |62| D2 | degrade to model tier + reasoning depth, never fake parallelism | no capability lever for the target tier on this host |63| D3 | fall back to a numbered text menu | host has no question tool |64| D4 | fixed at Light, no tier question, no sub-agents | Hook Context (auto-triggered) |6566This ledger deliberately diverges from skill-authoring.md §3b's column-or-separate-ledger rule for lane-applicability — `gold-standard` keeps a lane column, but here a column asserting a lane value that contradicted its own row's condition text measured worse than no column at all (`SKILL-VARIANCE-WALK.md` §Run 43: a bimodal Hook-Q4 split, the pre-registered key matching neither camp). D2 is restated at four sites in the shared partials — the general clause, the Standard row's "(else single-agent)", the Heavy row's "if supported", and the Heavy-specific "escalate by model + reasoning only" — all one row. D3 is stated once, in the shared Escalation footer's question-tool list ("…none → numbered text menu"). Neither is a new branch. D4's own branch text is restated verbatim in the footer's Hook Context line — same row, not a new one. The Freshness cap (scope already audited this session → cap at Light) is a tier-selection modifier on G1, not a degrade branch. The footer's Fix-mode-dependent offer clause is not a fifth branch — this skill defines no Fix mode section, so it never fires.6768## Output — 2 locations, declared69A location is a place this skill **writes** something a reader can see; the absence of an annotation is not one.70- Verified: `✅ [claim] — source: [link/file]`71- Unverified: `⚠️ unverified — check [exact source]`7273Stable fact: no annotation is written — not a location, not counted above.7475## AUTHORITATIVE vs DIVERSE76- **AUTHORITATIVE** (one ground truth): API/version/config/spec → go to the actual source code or official docs.77- **DIVERSE** (triangulate ≥ 3): "what's best" / landscape / patterns → multiple repos + docs + community; note conflicts.7879<!-- SHARED:ORCHESTRATION -->8081<!-- SHARED:ESCALATION_FOOTER -->
HetCreep