context-pack
Skill3 starsChanged 2 months agoMITView on GitHub
Prepare files, folders, code, logs, and structured data as a bounded, traceable context pack for Codex. Use when source material must be converted, deduplicated, selected, budgeted, or handed off with exact evidence anchors.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- richardochristjia/skillstheir own version, first seen Nov 1, 2026
- oocx/tfplan2mdtheir own version, first seen Oct 8, 2026
- Mesteriis/Engineering-Bible-AItheir own version, first seen Jun 28, 2026
- kambleakash0/agent-skillstheir own version, first seen Apr 6, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- oocx/tfplan2md★ 174same name
- kambleakash0/agent-skills★ 10same name
- richardochristjia/skills★ 2same name
- Mesteriis/Engineering-Bible-AI★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository6 other files it gives its agents.
- context-auditSkill
- context-benchmarkSkill
- context-onpremiseSkill
- context-orchestratorSkill
- conversation-checkpointSkill
- context-economySkill
The file
Files
Context Pack
Designed, integrated, independently refactored, and continuously maintained by TIKAZ.
Inputs and routing
Accept one or more files, folders, code trees, logs, structured data, or converter-produced Markdown plus a concrete task. Use text for confident text-first material, hybrid for bounded task-relevant visuals or complex tables, and source when extraction cannot preserve important evidence.
Workflow
Own canonical ingestion, fidelity profiling, exact deduplication, evidence selection, and final pack size. First run profile or let pack profile automatically:
text: canonical Markdown is sufficient;hybrid: use Markdown for text and a bounded visual-evidence queue for informative images or complex tables;source: keep the original asset/page path when safe extraction cannot preserve task-relevant information.
Do not trigger vision for a logo, repeated icon, background, or every image merely because it exists. When the queue contains pending-vision items and the host can inspect images, resolve the referenced item, record an anchored observation plus uncertainty, and keep the original reference. When the capability is unavailable, leave it pending or recommend the source file; never invent a description.
Profile first, protect literal facts and anchors, deduplicate only exact or formatting-only repetition, select task-relevant evidence, and assemble one task-ready artifact in this order:
- task and expected output;
- selected mode and estimated budget;
- confirmed constraints and protected facts;
- exact evidence excerpts with source anchors;
- decisions, completed work, and current state;
- conflicts and open questions;
- omitted-anchor inventory and verification limits.
The pack must distinguish exact source text, structured state, and inference. It must remain useful without the surrounding chat. Count the complete artifact against the budget. If essential protected evidence cannot fit, return a visible budget conflict instead of silently exceeding the limit.
Output contract
Return <directory>/packs/current-task.context.md with profile.json, visual-evidence.json, context-cost-ledger.json, canonical files, indexes, and savings-report.md. Distinguish exact excerpts, structured state, inference, omissions, and pending visual evidence.
Validation and fallback
Count the complete artifact against the budget and verify protected facts plus selected anchors. If a converter or vision host is unavailable, leave evidence pending or preserve the source reference. Never invent a visual description. If essential evidence cannot fit, report a visible budget conflict.
Example
Use context-pack on these release notes and logs. Build an 800-token pack for regression review, keep commands and versions exact, and list omitted anchors.Run python scripts/tikaz_context.py pack --input <path> --query <task> --budget <tokens> --visual-budget <items> --output <directory>.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.