embark

A command that unpacks a verified lifeboat—a portable bundle of project records—into a target code repository. It can restore architecture decisions, issues, intentions, specifications, and a marked section of CLAUDE.md.

In plain words
What is it for?
Probing or applying a lifeboat, reviewing unanswered coverage questions, and restoring selected project records from a previous repository state.
Why use it?
It lets developers inspect the proposed changes first and refuses unsafe input, conflicts, unknown write locations, or invalid files before changing the repository.

Command

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 commands/intentdriven/abcd/embark
Clone the repo
git clone --depth 1 https://github.com/intentdriven/abcd
Per session 64 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,034 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 $0.00064 $0.01034
Opus 5 $0.00032 $0.00517
Sonnet 5 $0.00013 $0.00207
Haiku 4.5 $0.00006 $0.00103

Measured yesterday against content hash 9e50dba07f37, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

embark 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.

commands/embark.md · 80 lines

How it starts

The opening of the file, as written. The whole thing — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/abcd:embark — unpack a lifeboat into a repo

Take a packed lifeboat (produced by /abcd:disembark) and write its record families back into a target repository. This is the inverse of disembark and the write half of the round-trip (adr-35): ADRs, issues, intents, and specs travel back verbatim, the current abcd marker block is re-injected into the target CLAUDE.md, and everything else in the lifeboat informs the report but is never written. The target defaults to the working directory when omitted.

Probe first (read-only)

Inspect what a lifeboat would write into a target without touching either:

"${CLAUDE_PLUGIN_ROOT}/abcd" embark probe <lifeboat-dir> [target-dir] --json

The lifeboat is untrusted input: probe gates it, verifies its manifest_sha256 against the on-disk tree, and refuses a symlink or an oversize file anywhere inside. Then it reports:

  • coverage — the brief blanks and their questions, surfaced FIRST. These are what a lifeboat hands a product thinker: the sections nothing in the record could ground, each with the question a human must answer (some marked human-owned).
  • planned — the record files that would land, each with its family, target_path, and action (create or, when the target already holds identical bytes, unchanged).
  • conflicts — target paths that would block a write (a differing file, a non-regular target, a non-directory parent). A plan with conflicts is still a successful probe.
  • ignored — lifeboat files not embarked (report-only, unmapped, unknown).
  • marker — what would happen to the target CLAUDE.md block.
  • record_manifest_sha256 — the record-derived closure seal.

Surface the coverage blanks to the human before anything else — they are the point of the handoff. Then summarise the plan.

Write

"${CLAUDE_PLUGIN_ROOT}/abcd" embark from <lifeboat-dir> [target-dir] --json

from runs the same planner as probe, then writes each create file into the target through two-layer containment (an os.Root boundary plus independent lexical validation), skipping unchanged files, and re-injects the current marker block into the target CLAUDE.mdnever foreign prose, only the canonical block. Summarise the result: written / unchanged, the per-families counts, and the marker action.

Read the full file on GitHub · 80 lines

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. yesterday First seen · 80 lines · 64 tokens per session scan A 9e50dba07f37

Subscribe to this mod's changes

embark is a command published in the GitHub repository intentdriven/abcd (3 stars, last pushed 2d ago), licensed MIT. It adds 64 tokens to every session and 1,034 once invoked, about $0.0003 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.

Related

Other commands, from other repositories