Add search-based skills: recall, replay, teach (#12) · Entire
Add search-based skills: recall, replay, teach (#12)
0fdf41f→main·
alishakawaguchi·1mo ago·9 files·+398 added/-6 removed
* search based skills
* Fix README merge conflicts and harden new skills
Resolve leftover merge markers in README and drop the code-tour block since that skill does not exist in this branch. Add shell-injection guardrails to recall, teach, and replay so user-supplied query text is passed as a single shell-quoted argument with metacharacters escaped. Remove stale code-tour cross-references from teach and replay.
Co-Authored-By: Claude Opus 4.7 (1M context) noreply@anthropic.com
* Use valid --date values and fix replay auth message
The entire CLI's --date flag only accepts week or month. Replace 90d/180d invocations with month, drop the 365d broaden step in favor of removing the filter entirely, and update the corresponding 90/180-day prose in guardrails and the README.
Also correct the replay skill's auth-required message, which referenced a non-existent entire replay subcommand instead of the actual gated command.
Co-Authored-By: Claude Opus 4.7 (1M context) noreply@anthropic.com
* Resolve replay step 4 vs step 5 contradiction
Step 4 told the agent to drop checkpoints whose transcripts cannot be read in step 5, but step 5 reads transcripts lazily one step at a time. Defer the unreadable-transcript handling to the existing inline skip rule in step 7's failure modes instead of asking step 4 to pre-fetch.
Co-Authored-By: Claude Opus 4.7 (1M context) noreply@anthropic.com
* Document recall, teach, replay, and session-to-skill in README
Co-Authored-By: Claude Opus 4.7 noreply@anthropic.com
* Revert README docs for recall, teach, replay, session-to-skill
These skills aren't ready to announce yet. Keep the skill definitions on the branch but pull them out of the public README.
Co-Authored-By: Claude Opus 4.8 (1M context) noreply@anthropic.com
* chore: bump version to 0.6.0
Adds the recall, teach, and replay skills.
Co-Authored-By: Claude Opus 4.8 (1M context) noreply@anthropic.com
Sessions
Transcript data is unavailable for this checkpoint.
Changes
9
.claude-plugin
Mmarketplace.json+1/-1
Mplugin.json+1/-1
.codex-plugin
Mplugin.json+1/-1
.cursor-plugin
Mplugin.json+1/-1
Mgemini-extension.json+1/-1
Mpackage.json+1/-1
skills
recall
ASKILL.md+123
replay
ASKILL.md+141
teach
ASKILL.md+128
{ "name": "entire", "description": "Cross-agent skills and commands powered by Entire", "version": "0.5.0", "version": "0.6.0", "source": "./", "category": "development", "author": {
M.claude-plugin/marketplace.json+1/-1
{
"name": "entire",
"version": "0.5.0",
"version": "0.6.0",
"description": "Cross-agent skills and commands powered by Entire",
"author": {
"name": "Entire Inc"
},
}
M.claude-plugin/plugin.json+1/-1
{
"name": "entire",
"version": "0.5.0",
"version": "0.6.0",
"description": "Cross-agent skills and commands powered by Entire.",
"author": {
"name": "Entire Inc",
},
}
M.codex-plugin/plugin.json+1/-1
{
"name": "entire",
"displayName": "Entire",
"description": "Cross-agent skills and commands powered by Entire.",
"version": "0.5.0",
"version": "0.6.0",
"author": {
"name": "Entire Inc"
},
}
M.cursor-plugin/plugin.json+1/-1
{
"name": "entire",
"description": "Cross-agent skills and commands powered by Entire.",
"version": "0.5.0",
"version": "0.6.0"
}
Mgemini-extension.json+1/-1
{
"name": "skills",
"version": "0.5.0",
"version": "0.6.0",
"type": "module",
"main": ".opencode/plugins/entire.js"
}
Mpackage.json+1/-1
---
name: recall
description: "Use when the user describes a task and wants to know whether something similar has been done before..."
---
# Entire Recall
Use `entire search` and `entire explain` to recall the closest prior session for a task and turn it into a playbook the user can act on. This is task-shaped, not list-shaped: the goal is "here's how to do your task" rather than "here are some checkpoints."
## Response Format
Begin the first response to this skill invocation with the line:
`Entire Recall:`
followed by a blank line, then the content.
- Apply the header to the **first response of the invocation only.** Do not re-print it on follow-up turns within the same invocation.
- Do **not** include the header on error...