claude

context-harness

Context management for Claude Code: fold completed tool results into one-line pointers at compaction instead of summarizing, keep the project's intent and status files current, and, with a Jev key, judgment gates for announced-but-undone turns, secret files under other providers, and conditional instructions. Folding needs no key; every Jev setting is off by default.

1 on GitHubby ysys143v0.0.3MITupdated 15 days ago

Not in a marketplace yet, so it loads for one session with --plugin-dir.

Plugin context-harness in ysys143/conjev. Commands read from the repo's manifest — 6 days ago.

What it installs

Skills (3)

  • consolidate

    Turn what a stretch of work produced into lasting memory - bring PHASE.md up to date, add episodes to the dated episode log, promote repeated procedures and decision-changing lessons into project skills, and propose lasting principles for AGENTS.md. Use when context-harness asks before folding, at t

  • project-state

    Keep INTENT.md (the agent's reading of what the person wants, sourced from their own words) and PHASE.md (the status board the person reads) current, alone or across several agent sessions. Use when work spans more than one session or compaction, when the person states or changes a goal, when severa

  • tune-policy

    Tune context-harness's folding policy with the person from measurements of their own session - measure, interview, dry-run a candidate, then apply it. Use when the person wants to change what context-harness folds, or says folding is too aggressive, too weak, or makes the model re-read too often.

View source on GitHub

More from ysys143/conjev

All 6 plugins →

conjev

claude

Context management for Claude Code: Jev judgment gates for announced-but-undone turns, secret files under other providers, conditional instructions, web pages and instruction files, around Claude Code's own compaction; folding completed tool results into one-line pointers instead of summarizing is there with fold: on. Keeps the project's intent and status files current. The Jev gates are auto by default and turn on once a key is in place and you consent (/jev-consent).

3 skills · 1 command

1

device-fixture

claude

Evaluator-only controlled stimuli for device acceptance.

1 command

1

chx

claude

Experiment: what is left in place of a folded tool result after compaction

1

device-sink

claude

Evaluator-only lower-chain fault and tool-kind fixtures.

1

long-task-observer

claude

Evaluator-only Stop observation and study-wide Jev spend accounting.

1

Similar plugins

discernment-nudge

claude

Appends two or three short follow-up questions after a substantive answer to help users check key facts, probe the reasoning, and notice missing context

19 skills

anthropics/skills 193,780

ruflo-ai-team

claude

Tenant-isolated AI teams that plan work, coordinate tasks, recall approved context, and return evidence-backed results

6 skills · 6 commands · 4 agents · 1 MCP server

ruvnet/ruflo 81,589

headroom-skill

claude

Skill: add headroom-ai compress() to a Python or TypeScript app to cut LLM input tokens (tool outputs, logs, JSON, chat history). No API key.

1 skill

headroomlabs-ai/headroom 79,508

context7

claudecodex

Up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your coding agent's context.

1 MCP server

upstash/context7 64,709

open-code-review

claude

Perform AI code review on Git diffs — supports workspace changes, branch ranges, and single commits with concurrent per-file analysis, codebase search, and deep context-aware review.

2 commands

alibaba/open-code-review 50,068

coderabbit

claude

Your code review partner. CodeRabbit provides external validation using a specialized AI architecture and 40+ integrated static analyzers—offering a different perspective that catches bugs, security vulnerabilities, logic errors, and edge cases. Context-aware analysis via AST parsing and codegraph relationships. Automatically incorporates CLAUDE.md and project coding guidelines into reviews. Useful after writing or modifying code, before commits, when implementing complex or security-sensitive logic, or when a second opinion would increase confidence in the changes. Returns specific findings with suggested fixes that can be applied immediately. Free to use.

2 skills · 1 command · 1 agent

anthropics/claude-plugins-official 47,076

Maintain context-harness?

This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.

claude & codex plugins6 plugins