jolliai/jolli-claude-plugin/recall
Claude Code and agent skill by jolliai/jolli-claude-plugin
- community
Checking that the SKILL.md file is still in the repository.
Recall prior development context for the current branch — decisions, why past choices were made, and where work left off. Use when the user wants to resume work, asks why something was done, or references earlier work, even without naming Jolli.
Install
npx vskill@latest install jolliai/jolli-claude-plugin/recallSKILL.md
Jolli Recall
Every commit deserves a Memory. Every memory deserves a Recall.
Load the structured development context for a branch — commits with their distilled topics (trigger / response / decisions / files), plus any plans and notes the work referenced — and synthesize a grounded answer to the user's prompt about that branch.
Step 1: Load the recall result
<user-arg> is a branch name (exact or fragment) or empty (current branch).
Preferred — MCP tool. If the recall tool from the jollimemory MCP server is available, call it with { "branch": "<user-arg>" } (omit branch when <user-arg> is empty). It returns a type-tagged object (recall / catalog / error).
Fallback — bundled CLI. If the MCP tool is unavailable, run the Jolli CLI through its stable dispatch script. Pass the argument via here-doc so it can never be interpreted as shell — do NOT interpolate <user-arg> into argv or a quoted string.
Generate a fresh random 16-character hex string (the "delimiter token") for this invocation — e.g. 3f8a9b2c5d7e1f4a — and replace the two <DELIM> occurrences below with it. Quickly scan the user's argument: if it contains a line that is exactly JOLLI_ARG_<delimiter token>_END, regenerate the token and re-check. A per-invocation random delimiter is what makes a pre-computed prompt-injection payload (a line that closes the here-doc early to smuggle shell) useless — do NOT hardcode a fixed delimiter.
Step 2: Handle the result by type
type:"recall"→ render the branch facts (branch,period,commitCount)
then walk commits[] (each has hash, commitMessage, recap?, and topics[] with always-present title + decisions). Surface decisions prominently; they are never dropped from a kept commit.
type:"catalog"→ semantic-match<user-arg>against the listed branches;
one match → repeat Step 1 with it, many → list and ask, none → show catalog.
type:"error"→ surfacemessageverbatim. For "no records": if the user is
not signed in to Jolli (see /jolli:status), note that memories are only generated once signed in and suggest /jolli:login (no Anthropic API key needed); otherwise suggest committing some work first so Jolli can generate memories.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingplugins/jolli/skills/recall/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/jolliai/jolli-claude-plugin/recall)