init-context

A project setup tool that adds a shared project direction, glossary, decision-record rules, and instructions for keeping them in context. It can also rebuild some of this information from the project's history.

In plain words
What is it for?
Use it to add project context files, a North Star describing who the project serves and what success means, a glossary, and architecture decision records. It can optionally seed these from existing commits and documents.
Why use it?
It prevents each coding session or teammate from using different terms, goals, or assumptions. It also gives new projects a useful starting structure when their past decisions are incomplete or missing.

Skill for Claude CodeCodex

Part of the manifest-dev plugin — 20 skills shipped together

Install

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.

agentmods
npx agentmods add skills/doodledood/manifest-dev/init-context
Any agent
npx skills add doodledood/manifest-dev --skill init-context
Clone the repo
git clone --depth 1 https://github.com/doodledood/manifest-dev

Made for: Claude Code, Codex.

Or install manifest-dev, the plugin that ships this one along with the rest of its 20 skills.

Per session 99 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,599 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5 $0.00099 $0.02599
Opus 5 $0.00049 $0.01300
Sonnet 5 $0.00020 $0.00520
Haiku 4.5 $0.00010 $0.00260

Measured 3d ago against content hash 533f6e815016, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

init-context 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 3d ago.

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.

claude-plugins/manifest-dev/skills/init-context/SKILL.md · 74 lines

How it starts

The opening of the file, as written. The whole thing — 74 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Install the project surfaces a repository needs so that every future session — and every teammate, with or without manifest-dev — works from the same direction, the same vocabulary, and the same decision record.

Two things get installed, and only the first is essential:

  • The wiring — an ADR conventions file the project owns, a glossary, a North Star (the project's standing strategy surface: who it is for, what it promises, what winning means), and a project-context-file section that makes sessions read and maintain all three. This runs on every project, including one with no history at all.
  • A seed — vocabulary and decision records reconstructed from the project's own history. This runs wherever there is history to read, and it is enrichment on top of the wiring, never a substitute for it.

Aim for a good starting point rather than a complete record. Most of a project's past reasoning is genuinely unrecoverable; a seed that is honest about what it could not recover is worth far more than one that reads complete and isn't.

Flags

--no-mine skips the seed and installs the wiring alone. Mining otherwise runs wherever there is history — a project with no commits, no docs and no code simply has nothing to seed from, which needs no flag to express.

Interpret only top-level options as flags; quoted or code-formatted mentions of them are topic text.

What already exists governs

Read before writing, and never overwrite a surface the project already maintains:

  • docs/adr/CONVENTIONS.md exists → the project has its own ADR convention. It governs. Do not replace it, and follow it for anything written this run — then compare it, per Compare, then offer below, which covers this file, docs/NORTH_STAR_CONVENTIONS.md, and the context file's surfaces section alike.
  • CONTEXT.md exists (or the context named by a root CONTEXT-MAP.md) → load it. Mined vocabulary is proposed as additions to it, never as a replacement.
  • NORTH_STAR.md exists → load it. Seeded content is proposed as additions or as evidence for positions it already carries — never as a rewrite: its positions change only by the owner's ruling, per the maintenance rules in references/NORTH_STAR_FORMAT.md.
  • docs/adr/ holds records but there is no CONVENTIONS.md → the project has an unwritten convention, and the records are the evidence of it. Infer it from them — naming scheme, which sections they carry, how they cross-reference, what the index looks like — and write the conventions file to describe that, adapting the shipped default rather than replacing it. Installing the default unchanged here would declare an existing corpus non-conforming, which is both wrong and the fastest way to get the file ignored. Where the corpus is inconsistent, or where following it would leave the index un-rebuildable, say so and let the user choose; never silently pick.
  • A project context file already carries some of this wiring → add the wiring it lacks and leave the rest alone; where its surfaces section is present but differs from the template, compare and offer rather than rewrite, per Compare, then offer below.

Read the full file on GitHub · 74 lines

Files

What ships with it

2 files 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.

Changes

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.

  1. 3d ago First seen · 74 lines · 99 tokens per session scan A 533f6e815016

Subscribe to this mod's changes

init-context is a skill published in the GitHub repository doodledood/manifest-dev (72 stars, last pushed 3d ago), licensed MIT. It adds 99 tokens to every session and 2,599 once invoked, about $0.0005 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.