MCPInstall

wheypoint

Mark the current conversation as a durable handoff so a new agent can resume the work. Use when the user wants to preserve state for a later or parallel session. Triggers include "hand this off", "write a handoff", "drop a wheypoint", "checkpoint this", "compact the conversation", and "/wheypoint". Also use for "wrap up" or "I need to clear context" during a task. Do NOT use for phase handoffs from /cook, /press, /age, or /cure.

Installs
620
Last 24h
—
Trending rank
—
Security
Not audited

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/paulnsorensen/easy-cheese --skill wheypoint
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads wheypoint automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

SKILL.md

The instructions your agent reads when the skill is triggered.

/wheypoint

/wheypoint records the state a new agent needs to resume the work.

Use it for culture sessions, for work without a phase slug, and for a phase's Checkpoint & stop option. A phase checkpoint is a standalone, non-terminal Wheypoint record, not a terminal phase artifact. Workers at a hard limit return observations; the parent, not the worker, runs checkpoint persistence.

Inputs

  • The conversation is the primary input.
  • The optional argument shapes only the orientation line; the focus never removes a decision, question, or directive.

Runtime commands

Run every command through this skill's archive by its resolved installed bundle path; see ../cheese/references/harness-portability.md. When the consumer repository lacks skills/, resolve the installed bundle path before execution; never assume the easy-cheese checkout or current working directory.

python3 skills/wheypoint/scripts/wheypoint.pyz turns [--session <id> | --transcript <path>]
python3 skills/wheypoint/scripts/wheypoint.pyz show <work-id> [--project <key>]
python3 skills/wheypoint/scripts/wheypoint.pyz validate [intent.json] [--base <revision_id>]
python3 skills/wheypoint/scripts/wheypoint.pyz checkpoint [intent.json] [--base <revision_id>]
python3 skills/wheypoint/scripts/wheypoint.pyz checkpoint <work-id> [--base <revision_id>] --question "<q>" --gates --rationale "<why>" --decision "<d>" --directive "<d>" --quote "<words>" --operator-directive "<style>" --operator-quote "<words>" [--owner <who> --when <trigger>] --link <ref> --kind <edge-kind> [--covers <id>]...
python3 skills/wheypoint/scripts/wheypoint.pyz schema checkpoint-intent
python3 skills/wheypoint/scripts/wheypoint.pyz resolve <absolute-path | work-id | slug> [--project <key>] [--workspace-root <checkout>]
python3 skills/wheypoint/scripts/wheypoint.pyz list [--scope project | machine] [--grep <text>]... [--status <s>]... [--next <move>]... [--project <key>]... [--entry-kind <k>]... [--entry-state <s>]... [--gated | --no-gated] [--edge-kind <k>]... [--linked-to <ref>]... [--forked-from <work-id>]... [--owner <who>]...
python3 skills/wheypoint/scripts/wheypoint.pyz log <work-id> [--project <key>]
python3 skills/wheypoint/scripts/wheypoint.pyz fork <parent> <child> --orientation "<child title>" [--next <move> [--artifact <a>]]
python3 skills/wheypoint/scripts/wheypoint.pyz export <slug | work-id | path> [--out <file>] [--include-operator] [--include-local]
python3 skills/wheypoint/scripts/wheypoint.pyz import <bundle.json> [--as <slug>] [--note-dir <dir> | --no-note]
python3 skills/wheypoint/scripts/wheypoint.pyz handoff render | parse | dispatch

Repeat --grep, --status, --next, or --project to search several values in one list call; a hit matching any value of one flag is kept, and distinct flags combine with AND; never loop over list once per term; outside a Git checkout, list with no --scope searches machine scope and reports scope-fallback: machine, while an explicit --scope keeps its meaning. fork writes only a child genesis record with one forked_from edge; import writes only an imported record, and bundle_digest covers the canonical bundle with no field of its own; references/work-graph.md gives refs, edge kinds, the bundle, and the list graph filters. Create a link with checkpoint --link; resolve validates artifact links, grounded paths, Git objects, and coverage; a slug with no exact match resolves by unique prefix, then unique substring, and several matches return ambiguous with every candidate. resolve, list, log, show, schema, and turns only read; direct invocations return output, and STOP before checkpoint writing. /cheese --continue uses resolve and never invokes another archive; slash commands are host renderings, not the control model; foreign machine hits require the owning checkout; follow the continuation protocol. The parent delegates persistence to this capability as one structured checkpoint task and runs validate before checkpoint; workers never invoke either command at a hard limit. Phase skills resolve with their own wheypoint-resolve --ref <slug> command; follow references/delta-contract.md for its outcomes, working_context, and findings.

Flow

  1. Read the user's words. Run turns and keep every user turn in view.
  2. Map each turn to an entry, or write one line that says why the turn is not captured.
  3. Rehydrate. Run show or resolve for the work id and note its revision_id; a first checkpoint needs no base.
  4. Every write to an existing record needs that revision_id as --base (or base_revision_id). After a compaction, read again before you write.
  5. Write the intent. Follow references/intent-contract.md.
  6. Put each user-stated constraint or preference in a directive entry with its verbatim quote.
  7. Put each choice in a decision entry with a rationale.
  8. Put each open item in a question entry; give a blocking question a rationale.
  9. Record a parked fork as a question with --gates and a --rationale that names the options and your prior leaning.
  10. Put the report a cold reader needs in notes.
  11. Put paths and URLs in artifact_links and working_context, not their contents.
  12. Validate. Run validate --base <revision_id> and fix every named problem.
  13. Checkpoint. Run checkpoint --base <revision_id>. A base-revision-required or stale-parent refusal means: read again, apply your change again, then write.
  14. Report. State the durability the result reports and the resume commands.

What the runtime enforces

The runtime refuses an intent instead of dropping data.

  • It refuses an unknown key and names its path.
  • It refuses a first checkpoint that carries no entry and no notes.
  • It refuses next: affinage without a PR reference in artifact.
  • It refuses next: cook without an artifact.
  • It refuses text that matches a credential pattern and names the field.
  • It refuses an empty artifact_links or remove_artifact_links list.
  • It derives status: from the gating entries per the handback contract; no author sets it.
  • It refuses a baseline key rather than drop it; a Cook baseline stays in the Cook handoff.
  • It requires a rationale for each blocking question a delta adds.
  • It derives every identifier, digest, and revision.
  • It refuses an agent-authored fork transition and a forked_from edge in a delta.
  • It refuses notes over 6000 characters.
  • It pins a bare path:line in working_context or an artifact link to path:line@<short HEAD sha> when HEAD holds the file, and otherwise warns with unpinned-line-ref (no checkout or commit, or an untracked, ignored, or absent file).
  • It lists each local reference (an xdg: ref, an absolute path outside the checkout, or a Git-ignored path) under ## Local references.
  • It warns with stale-notes when a decision changes and notes stay as they were, marks that drift at the top of ## Notes, and lists an entry that checkpoint --supersede <old> [--by <new>] --rationale <why> replaced under ## Superseded.

Fix a refused intent and run checkpoint again.

Handoff slug

The checkpoint command writes the shared preamble at the top of the generated projection; every consumer reads it with parse_handoff_slug().

status: <canonical status field>
next: mold | cook | press | age | cure | affinage | briesearch | culture | hold | tasks | done
artifact: <path, or PR#<n> / URL when next is affinage, else empty>
<one-line orientation: where the session is and what is mid-flight>

For next: tasks the projection adds a mode: parallel keyed line after artifact:; the keyed block after the orientation holds the Wheypoint pins. The projection body shows gates, open entries, decisions, directives, notes, context, artifacts, tasks, and a final ## Identifiers section. The projection title is # Wheypoint handoff. The projection is never the authority; never edit it and never resume from it by hand.

next: values

The meaning of each next: value is in references/intent-contract.md; briesearch and culture are read-only kickoffs, and tasks follows references/parallel-handoffs.md. A missing next: makes the handoff malformed; use hold when no action follows. Derive next: and status: from the open questions, not from expected success. Use status: gated: for every human decision; the resumed agent asks through the shared handoff gate before it dispatches. Handwritten notes, their legacy values, and their provenance fields are in references/legacy-notes.md and references/provenance-fields.md.

Rules

  • Never run a Git commit, push, or publication to raise durability.
  • Never write a note by hand, never edit a generated projection, and never include a secret value.
  • Reference each artifact by path or URL; do not copy its contents.
  • Use complete sentences; write one sentence per line.

Handoff

End with the orientation and this link: Wheypoint dropped: [.cheese/notes/<slug>.md](<absolute-note-path>).

FAQ

What does the wheypoint skill do?

Mark the current conversation as a durable handoff so a new agent can resume the work. Use when the user wants to preserve state for a later or parallel session. Triggers include "hand this off", "write a handoff", "drop a wheypoint", "checkpoint this", "compact the conversation", and "/wheypoint". Also use for "wrap up" or "I need to clear context" during a task. Do NOT use for phase handoffs from /cook, /press, /age, or /cure.

How do I install wheypoint?

Run npx skills add https://github.com/paulnsorensen/easy-cheese --skill wheypoint in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/wheypoint/ (personal) or .claude/skills/wheypoint/ (project).

Which AI agents can use wheypoint?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is wheypoint safe to install?

No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.

Who maintains wheypoint?

It is published by paulnsorensen in paulnsorensen/easy-cheese. It has 620 recorded installs.

More from paulnsorensen/easy-cheese

View all →

cook

paulnsorensen/easy-cheese

Implement an approved spec or focused task through stale-safe source edits. Use this skill when the user says "implement this", "cook this spec", or "fix this bug". Use it for /cook .cheese/specs/ .md. Also use it when the user says "go" with clear acceptance criteria. Run it alone for an unambiguous task. Use --prototype when the user says "prototype this", "go prototype it", or "spike this out" and wants to refine the result round by round. A spec helps but is not required. Do not use it for fuzzy planning (/mold). Do not use it for no-write discussion (/culture) or review-only work (/age).

mold

paulnsorensen/easy-cheese

Turn a high-leverage design decision into a validated spec through grounded dialogue. Use it when a leverage trigger fires (auth, irreversible effects, concurrency, contracts, destructive ops, a new slice, a cross-slice dependency, an invariant gap) or when the user asks for a spec or design. Typical phrases include "let's design X", "shape this into a spec", "what should the API for Z look like", and "/mold". A feature ask with zero fired triggers is a /cook mini-spec, not a mold; route it through /cheese. Do NOT use it for free-form discussion without artifact intent (/culture), direct implementation (/cook), or research-only questions (/briesearch). Ceremony scales to the ask through references/tiers.md, so a small clear change gets a validated mini-spec without a write-approval turn.

cure

paulnsorensen/easy-cheese

Apply selected fixes from an /age report, findings list, or CI failure. Run the project gates and hand a clean result to /plate. Use when the user wants selected findings resolved. Do not use for review, test authoring, or direct publication.

DevOps & Cloud649 installs

age

paulnsorensen/easy-cheese

Review a diff, PR, branch, or path across thirteen dimensions. Emit a severity-grouped findings report. Use when the user wants a code review. Trigger on "review this", "/age", "is this safe to merge", or "find bugs". Also trigger on "spot security issues", "check for slop", "review my PR", or "what's wrong with this code". Review every requested dimension. Review all thirteen dimensions by default. Do not apply fixes. Route them to /cure. Do not harden tests; route that work to /press.

cheese

paulnsorensen/easy-cheese

Route an idea, path, pull request, issue, failure, question, or bare /cheese to the correct workflow skill. Use this skill for /cheese, routing requests, help requests, or opening messages without a named workflow skill.

culture

paulnsorensen/easy-cheese

Models problems and trade-offs without writing code or specifications. Use when a workflow needs silent reasoning, or the user asks for "/culture", "no writes", "rubber-duck this", or "these bugs feel connected". Not for specs (/mold), code (/cook), or bug diagnosis and fixes (/pasteurize).

Popular code quality & workflow skills

View category →

grill-me

mattpocock/skills

A relentless interview to sharpen a plan or design.

Scan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick.

handoff

mattpocock/skills

Compact the current conversation into a handoff document for another agent to pick up.

domain-modeling

mattpocock/skills

Build and sharpen a project's domain model. Use when discussing codebase terminology, writing or editing a GLOSSARY.md, or recording or editing an ADR.

codebase-design

mattpocock/skills

Shared vocabulary for designing deep modules. Use when the user wants to design or improve a module's interface, find deepening opportunities, decide where a seam goes, make code more testable or AI-navigable, or when another skill needs the deep-module vocabulary.

diagnosing-bugs

mattpocock/skills

Diagnosis loop for hard bugs and performance regressions. Use when the user says "diagnose"/"debug this", or reports something broken/throwing/failing/slow.