Skip to content

fix(#8574)!: read .env as data in the OpenCode launch and runner probes - #8590

Merged
waynesun09 merged 4 commits into
mainfrom
agent/8574-opencode-env-as-data
Oct 20, 2026
Merged

waynesun09 merged 4 commits into
mainfrom
agent/8574-opencode-env-as-data

Conversation

@fullsend-ai-coder

@fullsend-ai-coder fullsend-ai-coder Bot commented Oct 19, 2026 •

Copy link
Copy Markdown
Contributor

Summary

Moves the OpenCode launch and the runner's in-sandbox probes onto envDataLoad. That is the pinned python3 -I -S loader #8565 added for Claude Code, codex and pi. With this change fullsend no longer sources the agent-writable /sandbox/workspace/.env or .env.d/*.env as shell anywhere it reads them.

Security-sensitive launch-path change. Human review is advisable.

Related Issue

#8574 (follow-up to #8498 / #8565)

Changes

  • OpenCode launch (internal/runtime/opencode_run.go): . .env is replaced with envDataLoad(envFile, openCodeEnvDataDrop...). It drops LD_*, PYTHON*, BASH_ENV, ENV, IFS and all 12 OpenCode read-only pins: the binary, utility and trusted-env/HOME/PATH pins. An assignment to a pin is dropped, so it can't abort the launch. The post-.env EnvExports re-pin, trusted-env restore, override reset and agent-file and adapter guards stay in place as a second layer. Comments that described .env as sourced are updated.
  • In-sandbox probes now use the same loader, so they see the values the launch sees and run no .env or .env.d line:
    • the pre-agent context scan (buildScanContextCommand) and the GitHub preflight (checkSandboxGitHubConnectivity), through a new exported runtime.EnvDataLoad. The preflight discards the loader's warnings. Otherwise a message like "line 401" would end up in the output that diagnosePreflightGitHubFailure searches for 401/404.
    • codex's harness hook-variable read (codexReadHarnessSecurityEnvCmd) and OpenCode's trusted-env read (openCodeTrustedEnvReadCommand).
  • Docs: the harness reference (.env.d host files), the security feature matrix row (OpenCode is now ✓, and the probes are mentioned), the OpenCode runtime page (the "still sourced" caveat is removed) and the customizing-agents guide.

Behaviour change (breaking)

OpenCode runs and the probes now take only [export ]NAME=value, export NAME and comment lines from .env/.env.d. Other lines are ignored with a warning that gives the file and line, never the content. Claude Code, codex and pi already behave this way. The codex harness-env read used to fail Bootstrap on a .env with a syntax error or a failing last line. It now skips that line and reads the values around it, as the codex launch does. A missing or unreadable .env still fails the read.

The launch needs python3 and a regular-file readlink in PATH directories the sandbox user neither owns nor can write (as for the other runtimes since #8591), and the context scan's find and the GitHub preflight's gh are looked up only in those directories. A gh placed under /sandbox/ (for example through host_files in /sandbox/workspace/bin) is no longer used by the preflight, which then reports gh CLI not available in sandbox and skips. The fullsend sandbox image is unaffected: gh and find live in /usr/bin.

Testing

  • New TestOpenCodeLaunch_HostileEnvDoesNotExecute, the issue's proposed test. It writes a .env and an .env.d file containing command substitution, backticks, a function redefining opencode, a bare command, exec, LD_PRELOAD/LD_LIBRARY_PATH/LD_AUDIT and assignments to the read-only pins, then runs the OpenCode prelude pieces against them. It checks that nothing runs, LD_* never reaches opencode, plain exports from both files still apply, PATH is restored to the pre-.env pin, the pinned binary starts, and warnings name the file and line but not the content.

  • The OpenCode run-command tests assert the rendered command contains the loader and no . <envFile>.

  • TestOpenCodeBinaryPin, TestOpenCodeTrustedEnvRestore and TestOpenCodeTrustedEnvReadCommand now go through the loader. The read test adds a marker command and a pin assignment.

  • TestCodexReadHarnessSecurityEnvCmd: the bad-line cases now expect the values to be read, a new case checks that shell constructs don't run, and the missing-file case still fails.

  • New shape tests for the scan-context command, both preflight commands, and EnvDataLoad. The scan command was also run in the sandbox image with a .env that puts another directory first on PATH: the image's find and the scanner run.

  • Rebased on main after fix(#8575)!: harden and tidy the .env-as-data launch loader #8591; go test ./internal/runtime/... ./internal/security/... ./internal/cli/... ./internal/harness/... pass as a non-root user on Linux (/bin/sh = dash).

  • go test ./internal/runtime/ ./internal/cli/ pass except five tests that fail identically on unmodified main in this sandbox, for environmental reasons: TestBuildRunCommand_ClaudeMissingExits127 (a real claude is on PATH), TestStartFetchService_* (sandbox networking), and TestPostScriptEnv_NoSchema* (the sandbox exports FULLSEND_OUTPUT_SCHEMA; they pass with it unset).

  • Patch coverage: every changed function is at 100%, except buildScanContextCommand at 89.5%; its uncovered line was already uncovered before this change.

  • pre-commit couldn't fetch hook repos in the sandbox (TLS error), so I ran the applicable hooks directly. gofmt, go vet ./..., the pre-commit-hooks checks (v6.0.0), lint-broken-symlinks, lint-docs-links and lychee all passed, and so did the gitleaks-based scan-secrets.

  • Tests added/updated for new or modified logic

  • make lint (pre-commit run) couldn't run in the sandbox; the hooks were run directly as described above

Checklist

  • PR title follows Conventional Commits (fix(#8574)!:, breaking-change marker plus BREAKING CHANGE: trailer)
  • Autonomous agent commit, so there is no DCO sign-off by design

🤖 Generated with Claude Code


Closes #8574

Post-script verification

  • Branch is not main/master (agent/8574-opencode-env-as-data)
  • Secret scan passed (gitleaks — 5fc4c47a1d40c2ab0a748e3bd2f7b9f8a697ca50..HEAD)
  • PR body secret scan passed (gitleaks — no-git)

@fullsend-ai-review

fullsend-ai-review Bot commented Oct 19, 2026 •

Copy link
Copy Markdown

🤖 Finished Review · ✅ Success · Started 9:43 PM UTC · Completed 9:58 PM UTC

Commit: 9fd166a · View workflow run →

Runtime: pi · Model: sonnet → claude-sonnet-5-5 · Effort: high · Cost: $2.87

@github-actions

github-actions Bot commented Oct 19, 2026 •

Copy link
Copy Markdown

Site preview

Preview: https://48288272-site.fullsend-ai.workers.dev

Commit: 3e0e0f8b049170efe294b73ee7dd10de4ef72dac

@codecov

codecov Bot commented Oct 19, 2026 •

Copy link
Copy Markdown

Codecov Report

❌ Patch coverage is 92.85714% with 1 line in your changes missing coverage. Please review.

Files with missing lines Patch % Lines
internal/runtime/opencode_run.go 50.00% 1 Missing ⚠️

📢 Thoughts on this report? Let us know!

@fullsend-ai-review fullsend-ai-review Bot added the risk/moderate PR risk: moderate label Oct 19, 2026
@fullsend-ai-review

fullsend-ai-review Bot commented Oct 19, 2026 •

Copy link
Copy Markdown

<!-- fullsend:risk-assessment -->
Risk Assessment: moderate (2/5)

Details

Moderate change size with no protected paths or dependency changes, balanced by elevated churn and multi-author contention in runtime packages alongside clean issue scope alignment for the unflagged launch behavior update.

Previous run

<!-- fullsend:risk-assessment -->
Risk Assessment: moderate (2/5)

Details

Moderate change size with no protected paths or dependency changes, balanced by elevated churn and multi-author contention in runtime packages alongside clean issue scope alignment for the unflagged launch behavior update.

Previous run (2)

<!-- fullsend:risk-assessment -->
Risk Assessment: moderate (2/5)

Details

Moderate change size with no protected paths or dependency changes, balanced by elevated churn and multi-author contention in runtime packages alongside clean issue scope alignment for the unflagged launch behavior update.

@fullsend-ai-review

fullsend-ai-review Bot commented Oct 19, 2026 •

Copy link
Copy Markdown

Review

Findings

Low

  • [injection-vuln] internal/cli/run.go:5711 — envDataDropNames (internal/runtime/env_data.go:77) lists only LD_PRELOAD, LD_LIBRARY_PATH and LD_AUDIT, although the commit message and docs say LD_* is dropped. The loader already supports prefix drops (PYTHON*), so export LD_TRACE_LOADED_OBJECTS=1 passes through EnvDataLoad. If a writer controlled .env, a glibc-linked find in buildScanContextCommand would print its library list and exit 0 without running the scanner, and the caller counts exit 0 as a pass. Exposure is limited: the probe runs once before the agent loop, pre_script is host-side, the host-side scan covers the same files, and the gap is unchanged from base. This PR reduces exposure compared with sourcing .env.
    Remediation: change the three LD_* names to LD_* in envDataDropNames and add a loader test that an exported LD_TRACE_LOADED_OBJECTS is dropped.

  • [breaking-config] docs/runtimes/opencode.md — (open, f_a9259fb34802e94a) This PR is a declared breaking change for OpenCode .env and .env.d files: only [export ]NAME=value, export NAME and comments are honored, and other lines are skipped with a warning. The OpenCode runtime page has no migration note for existing users. Remediation: add a short note to the .env is read as data bullet saying that shell logic must move into plain assignments.

Earlier findings

  • f_1983a5957285efd1 — resolved_by_change: docs/guides/user/customizing-agents.md now says launches and probes read .env.d as data and skip shell constructs with a warning.
  • f_efcd76ff4ea02263 — resolved_by_change: both "sourcing" mentions in docs/contributing/runtime-implementation.md now read "reading".
Previous run

<!-- fullsend:review-findings-v2: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 -->

Review

Findings

Medium

  • [injection-vuln] internal/cli/run.go:5710 — The pre-agent context scan now reads .env as data, but it still runs a bare find (and the GitHub preflight still runs a bare gh) under whatever PATH the loaded .env leaves behind. The loader keeps [export ]PATH=... lines, so an agent-written export PATH=... after the runner's PATH-restore control line decides which find/gh runs, and a planted find could exit 0 without calling the scanner. The new comment in run.go ("none can stand in for find or the scan") overstates this; only the scanner is pinned by absolute path. This is not a regression: the base code ran . .env as shell, which already allowed more. The PR narrows that exposure but leaves PATH-based executable selection open. Advisory, for maintainers to decide as follow-up.
    Remediation: Run find and gh by absolute paths resolved before .env is read (or reset PATH to the trusted value after EnvDataLoad in the probes), correct the run.go comment, and add a test with a hostile PATH line and a planted find/gh.
  • [incorrect-doc] docs/guides/user/customizing-agents.md:446 — Line 446 still says ".env.d files run as shell in the agent launch shell; treat them like pre_script." The PR edited line 431 of the same section to say Claude, codex, pi and OpenCode launches read these files as data, so the two passages now contradict each other. The PATH-restore sentences (around lines 372-374 and 442-444) also name only codex as restoring the pre-.env PATH, but OpenCode does too (FULLSEND_OPENCODE_PATH, opencode_run.go).
    Remediation: Reword line 446 to match docs/reference/harness-reference.md (launches and probes read .env.d as data, but .env is still valid shell an agent may source, so keep treating these files like pre_script). Add OpenCode next to codex in the PATH-restore sentences.

Low

  • [stale-doc] docs/contributing/runtime-implementation.md:163 — The 'OpenCode runner-held environment' bullet (around lines 162-163) and the 'OpenCode agent permissions and reset' bullet (line 172) still say 'after sourcing the agent-writable .env'. The PR changed OpenCode to read .env as data and updated the feature-matrix row and code comments, but not these two bullets.
    Remediation: Change 'sourcing' to 'reading' in both bullets so the wording matches the matrix row and opencode_run.go.
  • [incomplete-doc] docs/reference/harness-reference.md:185 — The host_files[].dest paragraph (line 185) says only Codex runs restore the pre-.env PATH. The PR's own .env.d paragraph (line 187) says OpenCode runs do too, and opencode_run.go confirms it (FULLSEND_OPENCODE_PATH is pinned and restored), so /sandbox/workspace/bin and .env.d additions are not on OpenCode's PATH either.
    Remediation: Change line 185 to say both Codex and OpenCode runs restore the PATH from before .env.
  • [breaking-config] docs/runtimes/opencode.md — The rules for which .env/.env.d lines take effect changed for OpenCode launches and the runner probes. Only [export ]NAME=value, export NAME and comment lines are taken; any other line is skipped with a warning naming the file and line. The OpenCode image now also needs python3 and a regular-file readlink in directories the agent cannot write, or the launch exits 127. The PR carries !, and the change is documented in harness-reference.md and opencode.md (with an exit-127 troubleshooting row), but docs/guides/ has no migration note for existing .env.d users.
    Remediation: Optional: add a short migration note to docs/guides/user/customizing-agents.md saying shell constructs in .env.d files are now skipped with a warning.

Next steps:

  • /fs-fix — agent addresses review findings automatically
  • /fs-fix <your instruction> — agent fixes with your specific guidance
  • Push commits directly — review re-runs automatically on push
  • /fs-fix-stop — disable automatic fix runs for this PR
Previous run (2)

<!-- fullsend:review-findings-v2:eyJ2ZXJzaW9uIjoyLCJhY3Rpb24iOiJhcHByb3ZlIiwiZmluZGluZ3MiOlt7InNldmVyaXR5IjoibG93IiwiY2F0ZWdvcnkiOiJicmVha2luZy1jb25maWciLCJmaWxlIjoiZG9jcy9ydW50aW1lcy9vcGVuY29kZS5tZCIsImlkIjoiZl9hOTI1OWZiMzQ4MDJlOTRhIiwiYWN0aW9uYWJsZSI6ZmFsc2V9XX0= -->

Review

Findings

Low

  • [breaking-config] docs/runtimes/opencode.md — OpenCode launches and the runner's in-sandbox probes (pre-agent context scan, GitHub preflight, codex and OpenCode env reads) now read .env and .env.d/*.env as data. Only [export ]NAME=value, export NAME and comments are taken. Other lines (command substitution, ;-joined commands, functions, unset, readonly, .) and assignments to LD_, PYTHON, BASH_ENV, ENV, IFS and the OpenCode pins are dropped with a warning. There is no opt-out or warn-only transition, so downstream .env.d host files that relied on shell semantics lose those lines at run time. The change is properly marked (! in the title, BREAKING CHANGE: trailer) and the docs are updated. The probes now also depend on python3. If it is missing, the loader pin fails and the GitHub preflight reports "GH_TOKEN not set in sandbox", which hides the real cause. The old . .env 2>/dev/null form fell through the same way, so only the python3 cause is new.
    Remediation: List the dropped constructs in the BREAKING CHANGE: release-note text. Consider giving the preflight probe a distinct skip reason when the loader fails.

Labels: PR changes Go runner and runtime env handling for agent launches and has security implications.

fullsend-ai-review[bot]

This comment was marked as outdated.

@fullsend-ai-review fullsend-ai-review Bot added ready-for-merge All reviewers approved — ready to merge go Pull requests that update go code component/runner Agent runner behavior and lifecycle security Security threat model and related concerns labels Oct 19, 2026
fullsend-ai-coder Bot and others added 2 commits October 20, 2026 02:01
envDataLoad, but the OpenCode launch still sourced the agent-writable
/sandbox/workspace/.env, so a line there or in an .env.d file could run
before the launch checks finished. The runner's in-sandbox probes also
sourced it as shell.

The OpenCode launch now reads .env through envDataLoad, dropping LD_*,
PYTHON*, BASH_ENV, ENV, IFS and every OpenCode read-only pin (the
binary, utility and trusted-env pins), so an assignment to one cannot
abort the launch. The post-.env re-pin, trusted-env restore, override
reset and agent-file guard stay as a second layer.

The probes use the same loader, so they see what the launch sees and
run no .env or .env.d line: the pre-agent context scan and the GitHub
preflight (through a new exported runtime.EnvDataLoad), codex's harness
hook-variable read and OpenCode's trusted-env read. The preflight
discards the loader's warnings so they cannot reach its 401/404
classification. The codex read now skips a bad line, as the launch
does, instead of failing; an unreadable .env still fails it.

The harness reference, the security feature matrix row, the OpenCode
runtime page and the customizing-agents guide are updated.

BREAKING CHANGE: OpenCode launches and the runner's probes now take
only `[export ]NAME=value`, `export NAME` and comment lines from .env
and .env.d files; any other line is ignored with a warning. Rewrite
host_files .env.d content that relied on shell constructs as plain
assignments, as Claude Code, codex and pi already require.

Closes #8574

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
OpenCode launches now read .env with the pinned python3, so the image
needs the same python3 and regular-file readlink the other runtimes do.

Assisted-by: Claude
Signed-off-by: Wayne Sun <gsun@redhat.com>
@fullsend-ai-review

fullsend-ai-review Bot commented Oct 20, 2026 •

Copy link
Copy Markdown

🤖 Finished Review · ✅ Success · Started 6:05 AM UTC · Completed 6:19 AM UTC

Commit: b11c094 · View workflow run →

Runtime: pi · Model: sonnet → claude-sonnet-5-5 · Effort: high · Cost: $3.40

fullsend-ai-review[bot]

This comment was marked as outdated.

fullsend-ai-review[bot]

This comment was marked as outdated.

@fullsend-ai-review
fullsend-ai-review Bot dismissed their stale review October 20, 2026 06:19

Superseded by updated review

@fullsend-ai-review fullsend-ai-review Bot removed the ready-for-merge All reviewers approved — ready to merge label Oct 20, 2026
@fullsend-ai-coder

Copy link
Copy Markdown
Contributor Author

🔧 Fix agent — iteration 1 (bot-triggered)

Fixed the four documentation findings and corrected the overstated run.go comment. Deferred the advisory PATH-pinning code change for find and gh as maintainer follow-up. The changes are docs plus one comment. gofmt and go build pass. Pre-commit could not run because hook repos were unreachable, and the full test suite was not run.

Fixed (5):

  1. run.go comment overstates that no .env line can stand in for find (internal/cli/run.go): Corrected the comment in buildScanContextCommand: a PATH assignment in .env is still taken, so find is resolved under that PATH and only the scanner is pinned by absolute path.
  2. customizing-agents.md still says .env.d files run as shell, contradicting the data-read wording; OpenCode missing from the PATH-restore sentences (docs/guides/user/customizing-agents.md): Reworded the paragraph to say launches and probes read .env.d as data and skip shell constructs with a warning, while .env stays valid shell, so the files should still be treated like pre_script. Added OpenCode next to codex in both PATH-restore sentences.
  3. runtime-implementation.md OpenCode bullets still say 'sourcing' the agent-writable .env (docs/contributing/runtime-implementation.md): Changed 'sourcing' to 'reading' in the OpenCode runner-held environment bullet and the OpenCode agent permissions and reset bullet.
  4. harness-reference.md host_files[].dest paragraph says only Codex restores the pre-.env PATH (docs/reference/harness-reference.md): Changed the paragraph to say Codex and OpenCode runs restore the PATH from before .env.
  5. No migration note for existing .env.d users about the breaking change (docs/guides/user/customizing-agents.md): The optional migration note was folded into the reworded .env.d paragraph, which now says shell constructs in .env.d files are skipped with a warning.

Disagreed (1):

  1. Pre-agent scan runs bare find and the GitHub preflight runs bare gh under a PATH that .env can set: The review marks this advisory, not a regression and not actionable, and leaves it to maintainers as a follow-up. Pinning find and gh by absolute path, or resetting PATH after EnvDataLoad, changes runner command construction and needs its own hostile-PATH tests. That goes beyond this PR's scope, so only the misleading comment was corrected.

Tests: passed

CI inspections (5):

  1. test (passing, success): Unit tests passed on the PR head.
  2. commit-lint (passing, success): Commit lint passed.
  3. behaviour (pending, in_progress): E2E behaviour job was still running when inspected.
  4. playback (passing, success): E2E playback passed.
  5. codecov/patch (passing, success): Patch coverage gate passed.

Conflict update: skipped; strategy none; forge MERGEABLE; target main

Next steps:

  • /fs-review — request a re-review of the changes
  • /fs-fix <your instruction> — run another fix pass with specific guidance
  • Push commits directly — review re-runs automatically on push
    Updated by fullsend fix agent

@waynesun09

Copy link
Copy Markdown
Member

/fs-fix-stop

@github-actions github-actions Bot added the fullsend-no-fix Skip bot-triggered fix agent runs label Oct 20, 2026
@github-actions

Copy link
Copy Markdown

Fix agent disabled for this PR. Remove the fullsend-no-fix label or use /fs-fix to re-engage.

@fullsend-ai-review

fullsend-ai-review Bot commented Oct 20, 2026 •

Copy link
Copy Markdown

🤖 Review · ⚠️ Cancelled · Started 6:28 AM UTC · Ended 6:34 AM UTC

Commit: 5b45c93 · View workflow run →

The pre-agent context scan and the GitHub preflight read .env as data,
but still ran a bare find and gh, looked up on the PATH .env left. Run
both with the trusted PATH recorded before .env (EnvDataTrustedPathVar).

Assisted-by: Claude
Signed-off-by: Wayne Sun <gsun@redhat.com>
@fullsend-ai-review

fullsend-ai-review Bot commented Oct 20, 2026 •

Copy link
Copy Markdown

🤖 Finished Review · ✅ Success · Started 6:36 AM UTC · Completed 6:44 AM UTC

Commit: 3e0e0f8 · View workflow run →

Runtime: pi · Model: sonnet → claude-sonnet-5-5 · Effort: high · Cost: $1.73

Comment thread internal/cli/run.go Resolved

@waynesun09 waynesun09 left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Rebased on #8591 and validated on Linux (non-root, dash) and in the sandbox image; bot approved 3e0e0f8, independent review found no MEDIUM+. LD_* prefix follow-up in #8607.

@waynesun09
waynesun09 added this pull request to the merge queue Oct 20, 2026
@github-merge-queue
github-merge-queue Bot removed this pull request from the merge queue due to failed status checks Oct 20, 2026
@waynesun09
waynesun09 added this pull request to the merge queue Oct 20, 2026
Merged via the queue into main with commit be8351f Oct 20, 2026
88 of 91 checks passed
@waynesun09
waynesun09 deleted the agent/8574-opencode-env-as-data branch October 20, 2026 09:40

This branch was successfully deployed

1 active and 1 inactive deployments
dev — 3e0e0f8b Deployed Oct 20, 2026 by waynesun09 via behaviour #17263
site-preview — 3e0e0f8b Deployed Oct 20, 2026 by github-actions[bot]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

component/runner Agent runner behavior and lifecycle fullsend-no-fix Skip bot-triggered fix agent runs go Pull requests that update go code ready-for-merge All reviewers approved — ready to merge ready-for-review Triggers review agent dispatch risk/moderate PR risk: moderate security Security threat model and related concerns

Projects

None yet

Development

Successfully merging this pull request may close these issues.

OpenCode launch and in-sandbox probes still source the agent-writable .env as shell

1 participant