Automatically recover working context after session compaction or when continuation is implied but context is missing. Works across Discord, Slack, Telegram, Signal, and other supported channels.
SkillScore73Maintained
Automatic
◆AI Engineer pack
Last verified 8/19/2026
What it changes
Passed 15 of 20 test cases with the skill, and 5 of 20 without it.
Without skill27%
With skill73%
Checks it passed that the model alone missed
Pipes output to tail -50
Uses the cat >> redirection operator
Pipes the extracted text through tail -20
and 12 more in the full report
11 fixed5 passed either way5 still failing1 broke
Pass-rate lift
+45 pp
Tokens
+85%
Agent turns
0%
Response time
-53%
Model gemini-3.6-flashMeasured 8/19/2026v116 of 22 cases passed with skill · 20 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 11 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 11
Prompt
You ran an initial history fetch of 50 messages on Slack, but the timestamps show only 45 minutes of conversation elapsed. What is the incremental step size for fetching additional history, and what is the maximum total message threshold allowed?
What was checked, without and with the skill
Checked
Without skill
With skill
Specifies fetching an additional 50 messages in the expansion step
Passed
Passed
Enforces a hard limit threshold of 100 maximum messages
Missed
Passed
case-05 · gemini-3.6-flash · 8/19/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md239 linessingle file
Context Recovery Skill
Automatically recover working context after session compaction or when continuation is implied but context is missing. Works across Discord, Slack, Telegram, Signal, and other supported channels.
Triggers
Automatic Triggers
Session begins with a <summary> tag (compaction detected)
User message contains compaction indicators: "Summary unavailable", "context limits", "truncated"
Manual Triggers
User says "continue", "did this happen?", "where were we?", "what was I working on?"
User references "the project", "the PR", "the branch", "the issue" without specifying which
User implies prior work exists but context is unclear
User asks "do you remember...?" or "we were working on..."
Execution Protocol
Step 1: Detect Active Channel
Extract from runtime context:
channel — discord | slack | telegram | signal | etc.
channelId — the specific channel/conversation ID
threadId — for threaded conversations (Slack, Discord threads)