context-summary is a Claude.ai Personal Skill. Say "context summary" in a long conversation and it reads the chat and compresses it into a fixed seven-block markdown handoff: what the work is, where the code stands, which decisions are locked, which traps are live, what is still open, where the archive lives, and what to do next. Paste that handoff as the first message of a new conversation and you resume at full fidelity without re-explaining the project, re-deriving decisions, or rediscovering the gotcha that already cost you an hour. It is a boarding pass, not an archive.
+----------------------------------------------------------+
| |
| c o n t e x t - s u m m a r y v2.2.0 |
| |
| Seven blocks. One paste. Zero re-explaining. |
| |
| [#] [#] [#] [#] [#] [#] [#] |
| |
+----------------------------------------------------------+
What It Does
The skill turns the live conversation into one markdown document with a fixed shape. The shape is the point: because the seven blocks always appear in the same order with the same meaning, a fresh session knows exactly where to look for the branch state, the locked decisions, and the first action to take. Empty blocks are omitted and named in a ledger line at the top, so a missing block reads as deliberate rather than forgotten.
flowchart TD
H[Seven-block handoff]
H --> A[TASK STATE]
H --> B[CURRENT HEAD]
H --> C[DECISIONS LOCKED]
H --> D[ACTIVE GOTCHAS]
H --> E[OPEN QUESTIONS]
H --> F[ARCHIVE POINTER]
H --> G[NEXT STEP]
Any of these phrases (or a close paraphrase) invokes it:
context summarycontext handoffgive me the handoffcarry context forwardhand this offprep the handoff
Real invocations look like this:
context summary
We are out of room here, give me the handoff.
Carry context forward, I am starting a fresh chat.
Demo
An abbreviated handoff from a multi-session Next.js API build, showing the shape of the output:
### TASK STATE
Building a multi-tenant SaaS API in Next.js. Story 1.1 (scaffold, base layout, route structure) done and pushed. Story 1.2 (AES-256-GCM credential encryption helpers, server-only) is next and sits on the critical path.
### CURRENT HEAD
Branch: main (primary integration branch, no feature branches in use for MVP)
Commit: 4bc7e2a security and config fixes after Story 1.1 review
### DECISIONS LOCKED
**Architecture:** Result<T> error wrapper at all external API boundaries; overrides generic try/catch (do not weaken, silent swallowed errors are the failure mode we are preventing).
### ACTIVE GOTCHAS
- AES-256-GCM requires a unique IV per encryption operation; symptom of IV reuse is silent encryption that is cryptographically broken with no runtime error; guard is generate the IV fresh inside every encrypt call, never pass it in.
### NEXT STEP
```
Continuing the B2B workflow automation API; resuming Story 1.2 (AES-256-GCM credential encryption helpers, server-only).
Begin Story 1.2 with manual review gates enabled.
The attached handoff is authoritative; read it before acting.
```
How It Works
You say a trigger phrase. Claude reads the conversation directly, with no copy-paste from you, and emits the seven blocks as markdown with no preamble and no commentary. If a git repo is in context, it pulls live branch and commit state instead of guessing. For non-code sessions, CURRENT HEAD adapts to describe the primary artifact and its state instead of git fields.
flowchart LR
A[Long session] --> B[Trigger phrase]
B --> C[Read conversation]
C --> D[Seven-block handoff]
D --> E[Paste in new chat]
E --> F[Resume work]
The handoff targets roughly 3,000 tokens and treats 5,000 as a hard ceiling; past that, archive material has leaked into the boarding pass and should be cut.
Installation
Claude.ai Personal Skills are installed through the Claude.ai settings interface. You need a Claude.ai account (Pro, Team, or Enterprise) that supports Personal Skills.
Step 1: Copy the Skill File
Download or copy the raw content of SKILL.md from this repository.
Direct link to raw file:
https://raw.githubusercontent.com/thebpandey/context-summary/main/SKILL.md
Step 2: Open Claude.ai Settings
- Go to claude.ai
- Click your profile icon (bottom-left of the sidebar)
- Select Settings
Step 3: Navigate to Skills
- In Settings, find the Skills section (also labeled "Plugins" in some interface versions)
- Select Personal skills
- Click Add skill or the equivalent button
Step 4: Paste the Skill
- Paste the full contents of
SKILL.md into the skill editor
- The skill name will auto-populate from the frontmatter as
context-summary
- Save the skill
Step 5: Verify Installation
Start a new conversation and type:
context summary
Claude should respond with the seven-block handoff structure immediately, without asking you to explain what it should do.
Troubleshooting
Skill does not trigger: Make sure the frontmatter at the top of the skill file was included when you pasted. The name: and description: fields in the frontmatter are what Claude uses to match trigger phrases.
Claude asks me to paste the conversation: The skill is designed to read the current conversation directly. If Claude asks you to paste, the skill may not have loaded correctly. Re-check that the full skill file content was saved.
Output is missing blocks: As of v2.2, empty blocks are omitted by design and listed in the Blocks empty this session: ledger line at the top of the handoff. A block absent from both the output and the ledger indicates an installation problem; confirm the full SKILL.md was pasted, frontmatter included. The version in this repo is v2.2.
Usage
At any point during a long working session, type:
context summary
Claude generates the seven-block handoff. Copy the entire output and paste it into the new conversation as your first message.
If you have notes, a prior handoff, or state from another conversation to fold in first, say:
I have context to add before you generate the handoff
Claude will pause for you to paste. After you paste, say continue and it will generate the handoff incorporating what you provided.
For short sessions (under roughly 10 turns, no accumulated decisions or gotchas), the skill produces a lean handoff: a brief TASK STATE, the populated blocks only, and a ledger line naming the omitted empty blocks. That is correct behavior, not a failure. Do not expect padded output for a thin session.
File Structure
context-summary/
SKILL.md the skill itself, paste this into Claude.ai
README.md this file
CHANGELOG.md version history
CONTRIBUTING.md how to propose changes
LICENSE MIT
.gitignore
assets/
hero.png banner image
Author
Bhaskar Pandey
Builder, technical strategist, and real estate professional. Not affiliated with Anthropic.
If you find issues or want to propose changes, GitHub Issues and pull requests are preferred. For everything else, reach out on LinkedIn or X.
License
MIT. Use freely, modify for your workflow, share with attribution.
