source-grounding

Source Grounding

Language: Generate EVERYTHING at runtime in the user's language — questions, answer options, menu labels, recommendations, report narrative. Detect from their messages; never default to English just because this file is English. English is allowed only for technical terms: commands, paths, code identifiers, severity labels (CRITICAL/HIGH/MEDIUM/LOW), and tier names (Light/Standard/Heavy).

Standing rule — active every response. No invocation needed for routine use.

What to verify (not memory)

• CRITICAL (always fetch or flag): 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

How

1. Identify the version-sensitive claim.
2. Name the authoritative source (official docs, advisory DB, package registry, spec, source code).
3. Fetch (WebSearch/WebFetch/docs MCP) — or flag ⚠️ unverified: check [source].
4. Cite at CRITICAL/MEDIUM. Don't over-verify stable facts.

Per-claim-type authoritative source map: read references/sources.md when choosing where to verify.

Source hierarchy

1. Source code / spec / RFC — primary ground truth
2. Official/vendor docs — authoritative secondary (honor .coalmine.json trustedDomains if set: treat those domains as additional authoritative / tier-2 sources)
3. Multiple reputable third-party sources — triangulated
4. Single blog — weak; corroborate first
5. Training memory — weakest for volatile facts

Non-interactive runs: log unfetchable claims as ⚠️ UNVERIFIED and continue — never block. Interactive: when sources cannot be fetched, confirm how to proceed via ask_question.

Output

• Verified: ✅ [claim] — source: [link/file]
• Unverified: ⚠️ unverified — check [exact source]
• Stable fact: no annotation needed

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.

Escalation — Scope & Model Quality

Tiers are capability targets, not platform commands — resolve each to your host's nearest lever. No lever for one? Degrade gracefully — never fake parallelism you can't do; escalate via model tier + reasoning depth instead.

Level	Intent	Capability target	Cost
Light	Spot-check key claims, single source	Cheapest model · single agent, no sub-agents.	Low
Standard	Balanced verification, mixed sources	Balanced model · raised reasoning · sub-agents per category only if your platform runs concurrent workers (else single-agent).	Balanced
Heavy	Full cross-verification, adversarial check	Most capable model + largest context · deepest reasoning · max sub-agent fan-out if supported · adversarial cross-check where available.	High

Per-platform Heavy levers + Heavy-run durability: read references/escalation.md before a Heavy run. No concurrent fan-out on your host → escalate by model + reasoning only.

Agent Context (interactive): score the tier rubric, then call ask_question once with the 3 tiers — the pick marked ✓, score shown, labels localized — and wait for the choice before starting. ask_question = your platform's question tool: Claude Code AskUserQuestion · Cline ask_question · Copilot askQuestions · Gemini CLI ask_user (business-tier product; individual tiers ended 2026-06-18 → Antigravity CLI) · Codex request_user_input · Cursor/Devin Desktop (ex-Windsurf)/Antigravity built-in prompts; none → numbered text menu.

Tier rubric (deterministic): +1 each — ① >20 files or whole-repo/cross-module reach ② >2 of this skill's categories relevant ③ release/security/pre-ship context ④ findings will drive code changes. 0–1 Light · 2–3 Standard · 4 Heavy. Freshness cap: scope already audited ≥Standard this session → cap at Light (re-auditing fresh ground wastes tokens; scope to what changed). Default tier: honor .coalmine.json defaultTier unless the user requests a tier for that run — an explicit request overrides everything.

Hook Context (auto-triggered): auto-Light, no tier question, no sub-agents — report first. Interactive session (a user is present) → offer the fix menu after the report; non-interactive → report-only. Never fix without a chosen option.

Entanglement: after the report, if confirmed findings fall in another canary's domain, offer it once via ask_question (one line, max one offer): perf/N+1 → scale-canary · contract/serialization/config → drift-canary · failure-path/retry → resilience-audit · logging/metrics → telemetry-canary · coupling/DI → testability-canary · dependency/CVE → supply-chain-audit · unverified version-sensitive claim → source-grounding · missing/stale rule → gold-standard.

Self error-report: if this skill misbehaves (contradictory instruction, broken procedure, wrong finding class), OFFER to file it at https://github.com/HetCreep/CoalMine/issues/new/choose with a user-reviewed summary — never auto-submit, never include unapproved code or paths.