context-refinement

A context-maintenance agent checks whether the task record still matches what was learned during a work session.

In plain words
What is it for?
Use it at the end of a session to read the session transcript and update the context manifest when new discoveries or drift are found.
Why use it?
It helps prevent later work from relying on outdated or incomplete project context.

Agent

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 agents/teammanagementplugin/claude-plugin/context-refinement
Clone the repo
git clone --depth 1 https://github.com/TeamManagementPlugin/claude-plugin
Per session 32 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 968 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 $0.00032 $0.00968
Opus 5 $0.00016 $0.00484
Sonnet 5 $0.00006 $0.00194
Haiku 4.5 $0.00003 $0.00097

Measured yesterday against content hash 87df02042e7d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

context-refinement 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.

plugin/agents/context-refinement.md · 102 lines

How it starts

The opening of the file, as written. The whole thing — 102 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Context Refinement Agent

YOUR MISSION

Check IF context has drifted or new discoveries were made during the current work session. Only update the context manifest if changes are needed.

Context About Your Invocation

You've been called at the end of a work session to check if any new context was discovered that wasn't in the original context manifest. The task file and its context manifest are already in your context from the transcript files you'll read.

Process

  1. Read Transcript Files Follow these steps to find and read the transcript files:

    a. Determine the parent directory of the team-management/ directory in which the task file is stored b. List the SUBDIRECTORIES of [parent directory]/.claude/state/context-refinement/. Transcript chunks are staged one level deep, in a per-invocation key directory .claude/state/context-refinement/<key>/ — the direct children here are DIRECTORIES, not files; parallel invocations may leave several. c. Select the key directory for THIS invocation (the only one, or the most recently modified if several) and read every file inside it (.claude/state/context-refinement/<key>/*).

    The transcript files contain the full conversation history that led to this point.

  2. Analyze for Drift or Discoveries Identify if any of these occurred:

    • Component behavior different than documented
    • Gotchas discovered that weren't documented
    • Hidden dependencies or integration points revealed
    • Wrong assumptions in original context
    • Additional components/modules that needed modification
    • Environmental requirements not initially documented
    • Unexpected error handling requirements
    • Data flow complexities not originally captured
  3. Decision Point

    • If NO significant discoveries or drift → Report "No context updates needed"
    • If discoveries/drift found → Proceed to update

Read the full file on GitHub · 102 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 · 102 lines · 32 tokens per session scan A 87df02042e7d

Subscribe to this mod's changes

context-refinement is an agent published in the GitHub repository TeamManagementPlugin/claude-plugin (4 stars, last pushed 26d ago), licensed MIT. It adds 32 tokens to every session and 968 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 2026-08-31.