Skip to content

test: fix rebrand-lag test failures; drop Windows installer - #2

Merged
s0ld13rr merged 1 commit into
mainfrom
fix/rebrand-test-lag
Jul 13, 2026
Merged

test: fix rebrand-lag test failures; drop Windows installer#2
s0ld13rr merged 1 commit into
mainfrom
fix/rebrand-test-lag

Conversation

@s0ld13rr

Copy link
Copy Markdown
Owner

The OpenCode→PentestCode rebrand updated product code but left test fixtures asserting the old branding, so bun turbo test and the HttpApi exerciser were red on main. This brings the fixtures in line with the rebranded behavior — no product-code changes, only test fixtures + Windows installer removal.

Rebrand-lag test fixes (36 unit failures + HttpApi auth)

  • provider plugins (vercel / openrouter / llmgateway / nvidia / kilo / zenmux / cerebras / cloudflare-*): expect pentestcode referer/title/source/integration/user-agent values instead of opencode.
  • webfetch: user-agent pentestcode.
  • observability: pentestcode.client / pentestcode.run resource attributes.
  • config: fixtures write pentestcode.json/.jsonc and .pentestcode dirs (the loader no longer reads opencode.*).
  • global / project: tmp dir pentestcode, project-id file .git/pentestcode.
  • agent / session-runner(-recorded) / permission: default agent is pentest (renamed from build).
  • location-layer: pentestcode.json fixture; drop apply_patch/edit/todowrite from expected tools (removed from core builtins in the fork).
  • skill: built-in skill is customize-pentestcode.
  • httpapi exerciser: send Basic auth user pentestcode to match the server's rebranded OPENCODE_SERVER_USERNAME default (was sending opencode → 401 on every protected route).

Legitimate opencode references were left intact: the opencode provider id, the opencode.ai/zen URL, app localStorage namespaces (opencode.global.dat) and DOM ids (#opencode-titlebar-right) — these are unchanged in product code, so e2e fixtures already match.

Windows installer removal

  • Delete install.ps1 (the Windows curl entrypoint) and script/sign-windows.ps1; no build/CI references remained.
  • Update the CLAUDE.md release checklist (no Windows targets / install.ps1).

Verified locally

  • packages/core tests: 1075 pass, 0 fail.
  • full bun turbo test: 8/8 tasks pass.
  • check:generated: clean.
  • test:httpapi coverage + auth: 208/208 pass.

The OpenCode→PentestCode rebrand updated product code but left test
fixtures asserting the old branding, so `bun turbo test` and the HttpApi
exerciser were red on main. Bring the fixtures in line with the rebranded
behavior (no product-code changes):

- provider plugins (vercel/openrouter/llmgateway/nvidia/kilo/zenmux/cerebras/
  cloudflare-*): expect pentestcode referer/title/source/integration/user-agent
  values instead of opencode.
- webfetch: user-agent "pentestcode".
- observability: pentestcode.client / pentestcode.run resource attributes.
- config: fixtures write pentestcode.json/.jsonc and .pentestcode dirs (the
  loader no longer reads opencode.*).
- global/project: tmp dir "pentestcode", project-id file .git/pentestcode.
- agent/session-runner(-recorded)/permission: default agent is "pentest"
  (renamed from "build").
- location-layer: pentestcode.json fixture; drop apply_patch/edit/todowrite
  from expected tools (removed from core builtins).
- skill: built-in skill is customize-pentestcode.
- httpapi exerciser: send Basic auth user "pentestcode" to match the server's
  rebranded OPENCODE_SERVER_USERNAME default.

Windows installer removal:
- delete install.ps1 (Windows curl-entrypoint) and script/sign-windows.ps1;
  no build/CI references remained. Update the CLAUDE.md release checklist.

Verified: core 1075/0, full `bun turbo test` 8/8, check:generated clean,
test:httpapi coverage+auth 208/208.
@s0ld13rr
s0ld13rr merged commit 684ae15 into main Jul 13, 2026
0 of 4 checks passed
@s0ld13rr
s0ld13rr deleted the fix/rebrand-test-lag branch July 13, 2026 19:59
s0ld13rr added a commit that referenced this pull request Jul 27, 2026
…ore-claim read-back

#2 (pentest.txt): judge a vector by SIGNAL, not the trophy. 'No root/marker yet after
25min' is NOT a kill reason — kill only on ZERO NEW SIGNAL across ~3 approaches (no
auth, no confirmed vuln, no accepted payload, no partial exec, no new cred). A vector
still producing signal is ALIVE — persist. A CONFIRMED-but-unfinished vuln → ESCALATE
to exploit_dev with the exact partial state (session/tokens/stored ids/missing step),
never kill. Fixes the cacti loss (real auth'd payload-accepted RCE killed at 27min
'because no marker'). Kill only for truly-dead; record via resolved_vectors.

#4 (kernel-exploit.txt): prove-before-claim — READ BACK, don't fire-and-forget. Before
recording access/root/marker as done, read it back through the same channel (ls/cat the
marker; confirm id=uid=0, don't infer root from a CVE name). MANDATORY on blind/OOB/
stateless channels where a write error is invisible. Can't confirm -> status suspected,
not achieved. Fixes the dubbo root miss (marker dropped blind, never read back, file
absent -> 0 credit) and prevents fabricated 'I have root' findings in real reports.
KumaloWilson pushed a commit to KumaloWilson/pentestcode that referenced this pull request Aug 2, 2026
When a versioned component has a public fixed release, diff patched vs target
version to locate the vuln from the fix, then focus exploitation on the changed
code. Target-agnostic (any versioned software) — no CVE/framework specifics, so
zero_day measurement stays clean. Depth-committing: also counters breadth-first
scatter. Complements read-the-handler (fix s0ld13rr#1).
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant