More screenshots
Works with the AI Agents you already use
Context Rot Forensics
Separate position loss from fill saturation in failing long-context AI sessions.
$19.99
Context Rot Forensics
Example session with this skill installed
Diagnose this 40-turn coding session. The database schema appeared near the beginning, the session is now about 75% full, and later output stopped respecting that schema. Evidence E3 confirms the schema was still present but missed. Distinguish position effects from fill saturation, propose only justified repair moves, and define an A/B rerun with pass criteria.
- Read your context and instructions
- Compiled the context rot forensics
Context Rot Diagnosis
Verdict: FILL SATURATION
Estimated fill: 75%
Missed item: db_schema
Observed zone: HEAD
Depth: approximately 15% from the start
Evidence cited: E3
The critical schema was not buried in the BODY zone; it remained in the HEAD. That weakens a pure position-rot explanation. The stronger supported diagnosis is saturation at high fill, with the missed schema becoming harder to use reliably as the session accumulated turns. This is an evidence-bounded classification, not a claim about hidden model internals.
Repair moves
O5 — Applied inference: create a fresh state summary that pulls the schema and current decisions into a compact TAIL position.
2.
O7 — Applied inference: reset the session with a structured handoff at roughly 75% fill instead of continuing to accumulate context.
A/B placement test
- A: rerun the next task with the original 40-turn context unchanged.
- B: rerun the same task with the schema and live decisions in a structured TAIL handoff.
- Keep model, prompt, temperature, and evaluation criteria constant.
- Pass condition: B follows every required schema field while A reproduces at least one omission.
If both variants pass or both fail, the repair is not proven and the verdict remains unresolved for that task.
Connects securely to your tools. The creator never sees your data.
About this skill
Diagnose whether a long-context failure is position rot, fill saturation, a mixed case, or not context rot. Context Rot Forensics maps critical information across HEAD, BODY, and TAIL zones, estimates context fill, cites the evidence behind the missed item, distinguishes observation from inference, and proposes a layout repair only when the diagnosis supports one.
The deliverable includes a scope-limited verdict, missed-item map, evidence ledger, repair moves, and an A/B placement test with exact pass criteria. The package includes the full SKILL.md workflow, CROT protocol, structured sample, deterministic context-map checker, tests, and a six-page PDF buyer guide.
It does not increase a model's context window, guarantee recall, summarize documents, or claim a repair worked without the supplied A/B result.
Does not increase a model's context window, guarantee recall, summarize documents, inspect hidden attention states, or prove a repair successful without the supplied A/B result. The verdict applies only to the session evidence provided.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 2 months ago
Permissions
No special permissions declared or detected