interlinked-cli

Coordinate with other agents/humans via the optional Interlinked MCP Server, and use local checkpoints & file reservations. Load this when picking up or handing off work (`tasks`, `send`, `inbox`, `handoff`), switching workspace context (`workspace`, `attach`), when a write is blocked because a file is "reserved by" another agent, when creating/restoring a local checkpoint (`checkpoint`, `rewind`, `resume`), installing the git-hook reservation guard (`interlinked guard`), or checking CI failure rates (`ci-status`). IMPORTANT: `checkpoint`/`rewind`/`resume` MUTATE git state (stash, checkout, clean -fd) and must not be run without explicit per-turn user authorization.

Understand and respond to the Interlinked PreToolUse guard — the local daemon that BLOCKS dangerous tool calls before they run. Load this when a Bash command or file edit was refused with "BLOCKED: … Suggestion: …", when you see an `[interlinked:<check>]` warning tagged `[proven]` or `[heuristic]`, when a destructive command / force-push / protected-file / secret / repo-confinement rule fired, when a grep was answered by the index, or when you need to know how to legitimately suppress a false positive or disable a guard rule. Covers what blocks, how to read the reason, suppression grammar, determinism tags, and the fail-closed cold fallback.

Inspect what AI agents did — the local, offline-first activity log and observability commands. Load this when you want to see what happened this session, tail activity live, review what the guard blocked/warned, find recurring mistakes, inspect a specific session/event, view the dependency graph, backfill external (Codex) sessions, verify the tamper-evident guard-decision log, or push buffered events to the server. Covers `status`, `activity`, `logs` (with `--type`/`--follow`), `explain`, `watch`, `telemetry`, `trace`, `search`, `recurrence`, `viz`, `audit`, `collect`, `compact`, `sync`, and the activity.jsonl / collection.jsonl / timeline.jsonl event-log model.

Respond to Interlinked's metric ratchets — the edit-time gates that block on line-count, cyclomatic complexity, coverage, and CRAP, plus the baseline-integrity gate. Load this when an edit was BLOCKED for growing a file past the line cap (500), for a cyclomatic-complexity jump or over-cap function, for adding an uncovered line or dropping coverage, for a high CRAP score, or for "loosening a baseline"; when adopting Interlinked on a legacy repo (`interlinked adopt`); or when managing caps/coverage/mutation baselines (`interlinked caps`, `coverage`, `mutation`, `debt`, `metrics`). The rule: water-lines only tighten — meet the bar (decompose / add a test / cover the line), never lower the baseline.

Install, operate, and troubleshoot the Interlinked CLI harness in a repo — install/uninstall agent hooks, connect runners (Claude Code, Codex, Copilot CLI, Gemini, Cursor), start/stop/restart the local guard daemon, run `interlinked doctor`, switch server / sync-mode / check-policy, log in, and manage the two-tier `.interlinked/` config. Load when setting up Interlinked, when `interlinked doctor` reports problems, when the guard daemon is down or stale, when hooks are not firing, or when configuring or disabling Interlinked.

Overview and router for the Interlinked CLI — a local guard, quality-enforcement, and observability layer for AI coding agents. Load this when working in a repo that has a `.interlinked/` directory, when you see any `[interlinked:*]` output or a `BLOCKED: … Suggestion: …` reason and are not sure which area it belongs to, or when you need to know what the `interlinked` command can do. This skill explains the mental model, the `.interlinked/` layout, the `[proven]`/`[heuristic]` tags, and routes you to the focused `interlinked-*` skill for setup, guard blocks, verify/checks, quality ratchets, supply-chain, spec-audit, observability, or coordination.

Keep prose specs and design docs honest against the code using Interlinked's spec-audit system. Load this when extracting a doc's invariants or facts (`interlinked spec invariants`, `spec agenda`), when you see `[interlinked:spec-drift]`, `[interlinked:spec-marker]`, `[interlinked:spec-xref]`, or `[interlinked:disputed-ground]` warnings after editing markdown, when ingesting/triaging an external review report (`interlinked findings ingest / status / verify / ack`), or when running executable doc examples (`interlinked doctest`). The system detects and tracks — it never auto-fixes; you write the fix. Covers the fact ledger, findings reconciliation, invariants/agenda, and current-state caveats.

Respond to blocked package installs and manage the Interlinked supply-chain allowlist. Load this when `npm/pnpm/yarn/bun/pip/pipx/poetry/uv/cargo/gem/bundle/go/composer/mvn/dotnet` install is BLOCKED with `[interlinked:supply-chain]`, when adding a dependency to package.json / requirements.txt / pyproject.toml / Cargo.toml / go.mod is refused, when a version is rejected as "not an exact pin", when `npm ci` fails a lockfile-snapshot check, or when you need to approve a package (`interlinked allowlist add`), snapshot a lockfile, or verify deps. Package installs are default-deny; a package needs both allowlist membership AND an exact version pin.

Run `interlinked verify`, understand the PostToolUse quality checks, and land multi-file edits through the content gate. Load this when you want to check your changes (`interlinked verify` — the on-demand whole-project check run), when a `pre_block` check refused an edit, when you need to land a cross-file refactor without transient tsc errors (`interlinked write --batch` / `multi-edit` / `verify-changeset` and the exporter-before-importers rule), when deciding whether a finding is default-gate or advisory, or when you need to know where to put probe/scratch scripts (`interlinked scratch`). Note: `interlinked verify` reports findings but exits 0 — it is not a pass/fail gate.

Distill imperative markdown guidance (AGENTS.md, AGENTS.md, .clinerules/, GEMINI.md, SKILL.md with hard imperatives) into Interlinked harness hook rules with verbatim source provenance. Invoke as /enforce <target> — local path, directory, GitHub shorthand (owner/repo/path), or URL — or no args to walk the project. Lexical strength is binding: never/MUST NOT/forbidden distill to block; should not/avoid to ask; should/prefer to advisory; hedged language is skipped. Output goes to .interlinked/distilled-rules.json plus .interlinked/distilled-rules.overrides.json. Lifecycle ops: /enforce list, show, remove, disable, enable, modify, add, reset, --review, --accept. Description-match invocation: make my AGENTS.md enforced, distill rules from this file. Manual invocation only — never auto-fires.

Distill imperative markdown guidance (AGENTS.md, CLAUDE.md, .clinerules/, GEMINI.md, SKILL.md with hard imperatives) into deterministic Interlinked harness hook rules with verbatim source provenance. Invoke as /enforce <target> — local path, directory, GitHub shorthand (owner/repo/path), or URL — or no args to walk the project. Lexical strength is binding: never/MUST NOT/forbidden distill to block; should not/avoid to ask; should/prefer to advisory; hedged language is skipped. Output goes to .interlinked/distilled-rules.json plus .interlinked/distilled-rules.overrides.json. Lifecycle ops: /enforce list, show, remove, disable, enable, modify, add, reset, --review, --accept. Description-match invocation: make my AGENTS.md enforced, distill rules from this file. Manual invocation only — never auto-fires.