agent/pi - pi - Penguin

Release v0.55.1

Mario Zechner · 2026-02-26 01:10:09 +01:00

v0.55.1 a753e24135

docs(changelog): cover upstream unreleased ai and coding-agent entries

Mario Zechner · 2026-02-26 01:13:32 +01:00

a27ea5b9e2

docs(changelog): audit unreleased entries and add missing items

Mario Zechner · 2026-02-26 01:11:34 +01:00

b2983c483e

chore: add repository .gitattributes for line ending normalization

Mario Zechner · 2026-02-26 01:11:34 +01:00

43c3c99dde

fix syntax for initializing AuthStorage in docs (#1539 )

Nader Dabit · 2026-02-26 00:35:13 +01:00

40a8706bb9

fix(ai): enable adaptive thinking for sonnet 4.6 and clamp xhigh effort (#1548 )

* fix(ai): enable adaptive thinking for sonnet 4.6 and clamp xhigh effort

* chore(ai): drop changelog edit from contribution

---------

Co-authored-by: tctev <224793535+tctev@users.noreply.github.com>

tctev · 2026-02-26 00:34:06 +01:00

e9d0074fa6

fix(ai): don't cache false for Vertex ADC credentials during async import race (#1550 )

`hasVertexAdcCredentials()` uses dynamic imports to load `node:fs`,
`node:os`, and `node:path` to avoid breaking browser/Vite builds. These
imports are fired eagerly but resolve asynchronously. If the function is
called during gateway startup before those promises resolve, `_existsSync`,
`_homedir`, and `_join` are still null — causing the function to cache
`false` permanently and never re-evaluate.

This means users with valid `GOOGLE_APPLICATION_CREDENTIALS`,
`GOOGLE_CLOUD_PROJECT`, and `GOOGLE_CLOUD_LOCATION` configured are silently
treated as unauthenticated for Vertex AI. Calls fall back to the AI Studio
endpoint (generativelanguage.googleapis.com) which has much stricter rate
limits, causing unexpected 429 errors even though Vertex credentials are
correctly configured.

Fix: in Node.js/Bun environments, return false without caching when the
async modules aren't loaded yet, so the next call retries. Only cache false
permanently in browser environments where `fs` is genuinely unavailable.

Co-authored-by: Jeremiah Gaylord <jeremiahgaylord-web@users.noreply.github.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>

jeremiahgaylord-web · 2026-02-26 00:31:16 +01:00

cf656c169c

fix(subagent): use getAgentDir for user agents path (#1559 )

Tianshu Wang · 2026-02-26 00:22:42 +01:00

7390f830de

feat(ai): add gemini-3.1-pro-preview to google-gemini-cli provider (#1599 )

Add Gemini 3.1 Pro Preview model to the Cloud Code Assist (google-gemini-cli)
provider for parity with the google and google-vertex providers that already
include this model.

Tested and confirmed working via the Cloud Code Assist API endpoint.

audichuang · 2026-02-26 00:18:03 +01:00

96cf7425ca

fix(coding-agent): handle typebox alias path separators on windows

Mario Zechner · 2026-02-26 00:28:23 +01:00

4939434997

fix(coding-agent): harden fd/rg bootstrap on windows

fixes #1348

Mario Zechner · 2026-02-26 00:19:43 +01:00

3db5715de4

docs(changelog): add unreleased coding-agent fix entries closes #1631 closes #1627

Mario Zechner · 2026-02-25 23:45:31 +01:00

ef8c950357

docs: skip generated model catalog changes in changelog audit prompt

Mario Zechner · 2026-02-25 23:42:57 +01:00

333ded7f26

fix(coding-agent): add offline startup mode and network timeouts (#1631 )

Matteo Collina · 2026-02-25 19:44:49 +01:00

757d36a41b

fix(tui): load koffi via createRequire in ESM (#1627 )

The previous lazy-loading change switched to require("koffi") inside
enableWindowsVTInput(). In ESM, require is undefined, so the call
threw and VT input mode was silently not enabled on Windows.

Use createRequire(import.meta.url) at module scope and
cjsRequire("koffi") at the call site. This keeps koffi externalized
for Bun binaries while restoring Windows VT input behavior, including
multiline paste handling.

herr kaste · 2026-02-25 12:05:27 +01:00

f129ac93c5

Add [Unreleased] section for next cycle

Mario Zechner · 2026-02-24 23:52:43 +01:00

5c0ec26c28

Release v0.55.0

Mario Zechner · 2026-02-24 23:52:01 +01:00

v0.55.0 d3232c2e9e

feat(coding-agent): prioritize project resources over global

Mario Zechner · 2026-02-24 23:50:55 +01:00

f0379384fe

Add [Unreleased] section for next cycle

Mario Zechner · 2026-02-23 01:57:06 +01:00

380236a003

Release v0.54.2

Mario Zechner · 2026-02-23 01:56:21 +01:00

v0.54.2 2417fc256f

fix(coding-agent): incremental highlight for streaming write tool calls

Mario Zechner · 2026-02-23 01:52:30 +01:00

0c61dd5873

docs(coding-agent): add changelog entry for extension theme persistence (closes #1483 )

Mario Zechner · 2026-02-23 01:03:25 +01:00

9e694f081a

Do not create empty .pi folder unconditionally (#1588 )

Duncan Ogilvie · 2026-02-23 00:59:10 +01:00

f1a2092bcf

fix(coding-agent): sync extension theme changes to session settings (#1483 )

Fero · 2026-02-23 00:57:36 +01:00

6137de9ce7

Extend OSS vacation to March 2, 2026

Mario Zechner · 2026-02-22 19:56:08 +01:00

34f841f036

docs(coding-agent): add built-in-tool-renderer to extensions README

Mario Zechner · 2026-02-22 19:05:57 +01:00

d34e8d80c4

fix(coding-agent): fix git-update test using unparseable git source

The test used 'github.com/test/extension' as the git source, but
parseGitUrl() requires a 'git:' prefix for bare hostnames. Changed
to 'git:github.com/test/extension' so the source is correctly
parsed as a git type and update() actually runs.

Mario Zechner · 2026-02-22 14:57:58 +01:00

316c2afe38

docs(coding-agent): add built-in tool renderer extension example

Shows how to override rendering of read, bash, edit, write tools
with compact output while delegating execution to the originals.

Mario Zechner · 2026-02-22 14:52:49 +01:00

21141e0040

fix(coding-agent): prefer provider/model split over gateway model id matching

When resolving --model zai/glm-5, the resolver now correctly interprets
'zai' as the provider and 'glm-5' as the model id, rather than matching
a vercel-ai-gateway model whose id is literally 'zai/glm-5'.

If the provider/model split fails to find a match, falls back to raw id
matching to still support OpenRouter-style ids like 'openai/gpt-4o:extended'.

Mario Zechner · 2026-02-22 14:40:36 +01:00

7364696ae6

Add [Unreleased] section for next cycle

Mario Zechner · 2026-02-22 14:35:28 +01:00

de1560a7ba

Release v0.54.1

Mario Zechner · 2026-02-22 14:34:44 +01:00

v0.54.1 5563d8969a

chore(ai): update generated models

Mario Zechner · 2026-02-22 14:33:51 +01:00

19c6f641cf

fix(tui): externalize koffi from bun binary builds

Changed koffi import from top-level to dynamic require in
enableWindowsVTInput() and added --external koffi to bun build.
This prevents embedding all 18 platform .node files (~74MB) into
every compiled binary. For Windows builds, only the win32_x64
koffi.node is shipped alongside the binary.

Binary size reduction: darwin-arm64 142MB -> 67MB, archive 43MB -> 28MB.

Mario Zechner · 2026-02-22 14:29:21 +01:00

8386a807ff

Add [Unreleased] section for next cycle

Mario Zechner · 2026-02-20 00:18:31 +01:00

3a3e37d390

Release v0.54.0

Mario Zechner · 2026-02-20 00:17:44 +01:00

v0.54.0 76b02a81b1

feat(coding-agent): discover skills in .agents paths by default

Mario Zechner · 2026-02-20 00:15:53 +01:00

39cbf47e42

docs(coding-agent): fix 0.53.1 changelog model provider summary