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.
git clone --depth 1 https://github.com/longthread/claude-pluginsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/longthread/claude-plugins/resume)<a href="https://agentmods.dev/commands/longthread/claude-plugins/resume"><img src="https://agentmods.dev/badge/commands/longthread/claude-plugins/resume/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/longthread/claude-plugins/resume"><img src="https://agentmods.dev/badge/commands/longthread/claude-plugins/resume.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00007 | $0.01008 |
| Opus 5.5 | $0.00003 | $0.00403 |
| Sonnet 5.5 | $0.00001 | $0.00202 |
| Haiku 4.5 | $0.00001 | $0.00101 |
Grade C, and why
resume scanned grade C with 1 finding 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 10d 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.
Tells the agent never to refusehighAnti-refusal
Suppressing the ability to decline removes a core safety control; a later harmful request then succeeds.
position, not after — a phase read without the arc around it is a phase you cannot refuse work How it starts
The opening of the file, as written. The whole thing — 81 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Resume the programme $1, or resolve it if $1 is empty.
Resolve the programme from the tree
git rev-parse --abbrev-ref HEAD
cat "<docsRoot>/programmes/INDEX.md"
Match the current branch against the branch column. $1 wins if given; otherwise
PROGRAMME_SLUG, when it names a programme directory, wins over the branch — it is how a relayed
session is told its programme. Skip any row whose status is closed — it is not a candidate,
even if its branch matches. If nothing matches and there is exactly one open programme, use it. If
several exist and none matches, list them and ask — do not guess, and do not fall back to "most
recently modified", which is how a session ends up writing into the wrong programme's ledger.
Several open rows on the current branch are the same case as none — list those rows and ask,
and take the answer as the user re-running /programme:resume <name>: the re-run is what pins the
programme for the rest of the session, including after a compaction.
Say you have started, before reading anything
As soon as the programme is resolved — before reading the ledger — print this as the first line of your first reply, in every mode, including autonomous:
Resumed programme "<SLUG>"
It signals that this session has resolved the programme and has started. A relaying predecessor waits for it in this pane's output, and the checks below can take many minutes, so it must not wait behind them. It says nothing of the checks, which come after.
Read, in this order
<docsRoot>/programmes/<SLUG>/ledger.md— in full, starting with## The arc: the goal, the programme's terminal condition, and which phases are stillplanned. Read it before the position, not after — a phase read without the arc around it is a phase you cannot refuse work inside.<docsRoot>/programmes/<SLUG>/NEXT-SESSION.md— read its "Start here" block now, but it is acted on only after the State-table checks and the report below: it is followed after the report, and in assisted and supervised mode after the user says proceed.
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.
- 10d ago Changed · +32 lines d252c71bd4b9
- 11d ago Changed · +2 lines 8202177a8c3a
- 1mo ago First seen · 47 lines · 7 tokens per session scan C 77a199060101
resume is a command published in the GitHub repository longthread/claude-plugins (1 stars, last pushed 11d ago), licensed MIT. It adds 7 tokens to every session and 1,008 once invoked, about $0.0000 per session on Opus 5.5. A static security scan graded it C with 1 finding (tells the agent never to refuse). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
learn
Codify implicit codebase knowledge as agent rules.
set-context
Set the client context so all subsequent analyses are automatically tailored — store client name, industry, size, geography, and key facts.
learn
Distill a finished conversation into classified knowledge candidates and propose (never auto-apply) memory writes or a new-skill hand-off. Closes the loop between a completed agent chain and the persistent memory system.
context
Analyze and optimize Claude Code's context configuration (analyze, optimize, or reset).
rust-teach
One-time setup that scans your Rust project, understands its patterns and conventions, and writes a Rust-specific context section to your CLAUDE.md. Run once per project to establish persistent Rust guidelines.
feature
Implement a ready-for-agent feature ticket in this checkout. Do not expand the ask.