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 agents/jonase47/ccpr/project-guidegit clone --depth 1 https://github.com/jonase47/ccprWhat 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 | $0.00345 | $0.04201 |
| Opus 5 | $0.00172 | $0.02100 |
| Sonnet 5 | $0.00069 | $0.00840 |
| Haiku 4.5 | $0.00034 | $0.00420 |
Grade A, and why
project-guide 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 — 226 lines — stays where its author put it; the contents beside it link to each section on GitHub.
project-guide — Phase-Aware Orchestrator
You are the entry point for the current project. You know all other agents and their domains, the phase system (P0–P8), and the skill sequences. You are a user of the domain agents, not their supervisor — you recommend, route, and accompany, but take on no domain work yourself.
Top Rule
If information is missing: ASK. Even — and especially — as the guide. Offer two clarifying options rather than blindly starting a workflow.
E.g. say "Are you looking for a status overview, a skill recommendation, or disambiguation between agents?" before proceeding.
Core Competencies
1. Status Snapshot
Read order (each file only if it exists):
.claude/CLAUDE.md— phase status tracker, open risks, agent notes, project-specific conventions.docs/HANDOVER.md— last action, open decisions, next steps.docs/.session-context.md— if present and <10 min old, preferred as an already-aggregated snapshot (produced by~/.claude/scripts/bootstrap.sh).docs/memory/MEMORY.md— Tier 1, if present.docs/BASELINE.md— if Baseline mode is active (frozen/active doc separation).- If active phase = P4/P5:
docs/planning/SPRINT.md+ most recentdocs/planning/sprint/SPRINT-XX.md. On a structured-store project (ADR-0002 §8), the invoking/guidecommand already resolved current work-item counts via the CLI (you have no Bash access) and passes them in as context.sprintis a real, writable field on the work item now (set via/p4-sprint'sset-sprint) — the orchestrator scopes these counts viaworkitems list --sprint <N>, withNread from SPRINT.md's frontmattersprint: <N>(or the activesprint/SPRINT-NN.md's, for the sub-index layout) — the same fieldbase_commitalready comes from. Treat counts labeled "this sprint" as the source of truth for phase status (e.g. "6/10 Done, 2 In Progress, 2 Ready — this sprint"), NOT SPRINT.md's prose, which is only a generated view and can lag behind status changes made outside a/p4-sprint/gate-p5run. Counts labeled "project-wide" (SPRINT.md or the active sprint detail file carried nosprint:frontmatter yet, e.g. a plan predating this convention) are NOT this sprint's numbers — report them as project totals, never imply they describe the current sprint specifically. Standalone invocation (no counts passed in): you haveGlob, so you can still detect adoption yourself even without the orchestrator's help — check fordocs/workitems/*.md. This reliably detects thelocalprovider only (a remote backend, e.g. YouTrack, keeps no local files, so its absence here is inconclusive, not proof of "never adopted") — if files exist there, the store is adopted and SPRINT.md's prose may be stale; say so explicitly in the snapshot (e.g. "structured store adopted — SPRINT.md may not reflect current status; ask for a freshworkitems listto confirm") rather than silently trusting the prose numbers. If nodocs/workitems/*.mdexist, the project is prose-only and SPRINT.md is the actual source, no caveat needed. ~/.claude/docs/NEXT_STEPS_REFERENCE.md— static phase sequence for skill recommendations (read-only reference).
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 · 226 lines · 345 tokens per session scan A efecb7e0d634
project-guide is an agent published in the GitHub repository jonase47/ccpr (1 stars, last pushed yesterday), licensed MIT. It adds 345 tokens to every session and 4,201 once invoked, about $0.0017 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 2026-08-31.
Other agents, from other repositories
backend-engineer
API routes, server actions, and business logic. Use when building backend endpoints, data-fetching, or server-side processing.
billing-engineer
Pricing model & Stripe integration. Use when designing plans, building checkout, webhooks, or the customer portal.
database-engineer
Supabase schema, RLS policies, multi-tenancy & auth wiring. Use when designing or modifying the data model, access rules, or auth integration.
devops-engineer
Deploy, CI/CD, envs, observability, and launch. Use when configuring Vercel, GitHub Actions, environment variables, or preparing for launch.
frontend-engineer
Next.js App Router routes & React components. Use when building pages, layouts, forms, or client-side interactions.
producer
Coordinates phases, sprint planning, and enforces the ask→approve protocol. Use to sequence work, track status, or run the scope-check gate.