Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/phase-rs/phase/engine-implementernpx skills add phase-rs/phase --skill engine-implementergit clone --depth 1 https://github.com/phase-rs/phaseWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00043 | $0.07683 |
| Opus 5 | $0.00022 | $0.03841 |
| Sonnet 5 | $0.00009 | $0.01537 |
| Haiku 4.5 | $0.00004 | $0.00768 |
Grade A, and why
engine-implementer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 232 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Engine Implementer (Orchestrator)
This is the orchestrator for the phase.rs implementation pipeline. It runs as a skill in the main thread so it can spawn agents for every step that benefits from fresh context (plan review, surgical implementation, implementation review). Do not turn this into an agent — agents cannot spawn sub-agents, which is what made earlier versions silently degrade.
Roles
| Step | Where it runs | Why |
|---|---|---|
| 1. Produce plan | Spawned general-purpose agent invoking /engine-planner |
Fresh context = plan is shaped by the task, not by the conversation history that led here |
| 2. Review plan | Spawned general-purpose agent invoking /review-engine-plan |
Fresh context = honest architectural review, independent of the planner |
| 3. Implement | Spawned engine-implementation-executor agent |
Baseline measurement, surgical edits, and preparatory checks; never commits |
| 4. Checkpoint + measure | This thread, then a fresh measurement executor | Orchestrator creates the candidate commit; isolated executor measures that immutable candidate |
| 5. Complete verification | This thread | Verify the committed candidate, never an in-flight working tree |
| 6. Review implementation | Spawned general-purpose agent invoking /review-impl |
Independent review of the immutable base-to-candidate diff |
| 7. Final acceptance | This thread | Accept only the exact reviewed checkpoint candidate |
Runtimes without subagent spawning (contributor environments — Codex CLI, plain LLM sessions). The pipeline's value comes from context isolation between author and reviewer, not from the spawning mechanism. If your runtime cannot spawn agents, do NOT silently degrade to reviewing your own work in the same context — that is the failure mode this skill exists to prevent. Instead: run each step against a fresh context (new session/conversation per step when your runtime supports it), and for every review step hand the reviewer ONLY the artifact under review (the full plan, or the unified diff), the original task description, CLAUDE.md, the relevant skill (/review-engine-plan or /review-impl), and, in chartered runs, the charter, phase index, and deferral allowlist — never the conversation that produced it. If even that is impossible, say so explicitly in the final report and in the PR body under a "Validation Failures" heading; do not claim the review loop ran clean.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 232 lines · 43 tokens per session scan A 1d0839b36197
engine-implementer is a skill published in the GitHub repository phase-rs/phase (259 stars, last pushed yesterday), licensed Apache-2.0. It adds 43 tokens to every session and 7,683 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
qa-cli-mcp-api
Select and run explicitly requested, risk-based QA for Anarlog's CLI, webhooks, stdio MCP, hosted Cloud API, and remote MCP. Test only affected lanes unless comprehensive coverage is requested.
product-update-newsletter
Draft, update, or audit a crisp, changelog-grounded Anarlog product-update newsletter in Loops (app.loops.so) for a desktop release. Use after the changelog is merged, when asked to draft, revise, or pre-send check the release announcement email.
sqlite-schema-design
Design or review schemas for crates/cloudsync using SQLite Sync constraints, not generic SQLite advice. Use when adding synced tables, changing synced columns, or planning CloudSync-safe migrations.
release-new-version
Release a new desktop stable version for Anarlog. Use when asked to cut, publish, or prepare a new stable desktop release after checking and merging the changelog.
fix-ready-prs
Inspect every open non-draft PR for CI failures and unresolved Cursor Bugbot findings, then fix them on the existing PR branches. Use when asked to check active PRs, triage ready PRs, fix CI, or address Bugbot comments.
reactive-sqlite-ui
Build SQLite-backed reactive UI in apps/desktop using stable patterns for reads, selection, forms, writes, and loading states. Use when implementing or reviewing screens built on useDrizzleLiveQuery and SQLite mutations.