codemode-output
Present native code-mode reads or edit receipts without double JSON escaping; recover partial output and cached read_many pages, or reread after a…
Skill by Magaav
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/magaav-codemode-output
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: writing
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/codemode-output
What's inside
- Skill name
codemode-output- Description
- Present native code-mode reads or edit receipts without double JSON escaping; recover partial output and cached read_many pages, or reread after a formatter/stale-edit refusal. Use when aggregate native output is large or receipts are noisy, not for trivial calls.
- License
- MIT
- Instructions
- 46 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# Native output and fresh receipts
Load `docs/CODEMODE.md` (Structured native presentation) for the versioned contract.
This is opt-in presentation, not a smaller global output budget or a new cache.
```js
await tools.read_many({requests:[{path:'docs/A.md',limit:120},{path:'docs/B.md',limit:120}]});
await showCall({version:1,call:1});
```
`call` is the one-based admitted native-tool call index, not a discovery/helper
attempt ID. `await describeTool('showCall')` gives the helper contract. Do not also
`text`/return the same large result: that adds escaped copies. Small counters or
explicitly selected diagnostics remain suitable for `text`.
Inspect `shown`, `presentation_sanitized`, `presentation_redacted`, per-item errors,
EOF/byte continuations and cached `next_cursor`. Displayed content is a normal
redacted model view, not byte-identical private evidence. Original source byte counts
and selections remain tied to the captured original. Never form old_text from masked
source. JSON-key redaction collisions refuse visibly, not complete empty success.
A visible captured page is not necessarily a whole document or whole batch.
Continue later cached results with ordinary `read_many({cursor:...})`; do not rerun
the script, particularly effects. If `presentation_partial`, inspect all unread
coverage before claiming review complete. Each unread range names the native
original SHA-256, exact offset and byte count. Retrieve with `tool_result`, following
its bounded `next_offset` until the range is covered. This boundary fallback costs
extra retrievals; retained bytes are not model awareness. No automatic pagination,
refetch, retry, arbitrary formatter inference or rollback occurs.
After a formatter or stale-edit refusal, explicitly reread current source, then
form a NEW exact edit from that source/selection. Do not alter old_text by guessing
or reuse the stale selection. A fresh receipt is a new snapshot, not recovery of
the old captured source. Keep successful earlier effects and unknown outcomes
visible; never replay the earlier script to obtain its output or fresh receipts.
```js
const current = await tools.read({path:'src/example.lua',offset:20,limit:12});
// Inspect current.content before constructing the next exact edit.
await showCall({version:1,call:1});
```
Then use ordinary `edit` with that exact current selection and explicit inclusive
line frame, or a uniquely occurring old_text copied from the current read. Separate
interpretation-dependent steps rather than inventing a mutating retry helper.From Magaav/wasm-agent, MIT.
More from Magaav/wasm-agent
- 0.592score
code-graph
Magaav
Retrieve a ranked code symbol and its exact implementation, answer explicit dependency questions, or audit a patch's possible missed callers.
SkillAFreshness: A★ 1Claude Code +9 - 0.588score
parallel-evolution
Magaav
How to change code on a branch while other people or agents may be changing the same repository, so your change stays current with the integration branch and…
SkillAFreshness: A★ 1Claude Code +9 - 0.588score
git-orchestrator
Magaav
The merge lane's runbook and the one home of the integration protocol: who may move main, the lifecycle (produce -> verify -> accept -> land -> deploy ->…
SkillAFreshness: A★ 1Claude Code +9 - 0.587score
history-search
Magaav
Retrieve original past requests, decisions, corrections or execution evidence efficiently.
SkillAFreshness: A★ 1Claude Code +9 - 0.587score
whatsapp-reply
Magaav
Scoped WhatsApp journal reads/decisions and linked-device transcription safety.
SkillAFreshness: A★ 1Claude Code +9 - 0.587score
client-controls
Magaav
How to act on the user's own machine (screenshot, mouse, keyboard, a shell) and how to browse a web page in Chrome.
SkillAFreshness: A★ 1Claude Code +9 - 0.585score
agent-benchmark
Magaav
Qualify isolated four-arm coding pilots, verify hidden acceptance and lossless cleanup, then compare only explicitly approved trials.
SkillAFreshness: A★ 1Claude Code +9 - 0.584score
gate-lifecycle
Magaav
Inspect a queued, stuck, cancelled or nested test gate; recover only after proving process/lease ownership and drain.
SkillAFreshness: A★ 1Claude Code +9 - 0.584score
delivery-recovery
Magaav
Reconcile concurrent delivery-record updates or an admission event whose enqueue failed, returned queued zero, or was interrupted.
SkillAFreshness: A★ 1Claude Code +9 - 0.583score
vm-windows
Magaav
Use when bootstrapping, starting, repairing, accessing by SSH, or testing wasm-agent in a disposable Windows Hyper-V VM on the operator's PC.
SkillAFreshness: A★ 1Claude Code +9 - 0.583score
clean-disk
Magaav
How to reclaim disk on this machine safely: measure first with cheap probes, delete only what this file names, never touch the node's state, the databases…
SkillAFreshness: A★ 1Claude Code +9 - 0.583score
self-update
Magaav
How to rebuild and replace the node you are running on - and, when the change is in the supervisor itself, how to deploy a new wa-sentinel - without breaking…
SkillAFreshness: A★ 1Claude Code +9
Signals
- Stars
- 1
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 16, 2026 (12 days ago)
- Last release
- Oct 20, 2026 (9 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/Magaav/wasm-agent
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from Magaav
- 0.592score
code-graph
Magaav
Retrieve a ranked code symbol and its exact implementation, answer explicit dependency questions, or audit a patch's possible missed callers.
SkillAFreshness: A★ 1Claude Code +9 - 0.588score
parallel-evolution
Magaav
How to change code on a branch while other people or agents may be changing the same repository, so your change stays current with the integration branch and…
SkillAFreshness: A★ 1Claude Code +9 - 0.588score
git-orchestrator
Magaav
The merge lane's runbook and the one home of the integration protocol: who may move main, the lifecycle (produce -> verify -> accept -> land -> deploy ->…
SkillAFreshness: A★ 1Claude Code +9 - 0.587score
client-controls
Magaav
How to act on the user's own machine (screenshot, mouse, keyboard, a shell) and how to browse a web page in Chrome.
SkillAFreshness: A★ 1Claude Code +9 - 0.587score
whatsapp-reply
Magaav
Scoped WhatsApp journal reads/decisions and linked-device transcription safety.
SkillAFreshness: A★ 1Claude Code +9 - 0.587score
history-search
Magaav
Retrieve original past requests, decisions, corrections or execution evidence efficiently.
SkillAFreshness: A★ 1Claude Code +9
Cite this page
codemode-output skill. Agent Mart. https://agentmart.dev/skills/magaav-codemode-output (last updated 2026-10-24)