fix(opencode): preserve tool context through compaction and prompt loops

[GuestAUser]

Issue for this PR

Closes #20246

Type of change

• Bug fix
• New feature
• Refactor / code improvement
• Documentation

What does this PR do?

This PR fixes three related session-context problems in packages/opencode/src/session.

• It preserves bounded tool evidence through compaction so follow-up reasoning can still reference meaningful earlier tool output.
• It threads OpenAI Responses tool follow-ups with previous_response_id when store: true is explicitly enabled, so the continuation path sends the tool-result delta instead of replaying the full prompt history.
• It memoizes the stable per-loop prompt fragments from SystemPrompt.skills(agent) and SystemPrompt.environment(model) while still reloading instruction.system() on every iteration so instruction changes remain visible on the next loop step.

Together, these changes make multi-step sessions more context-preserving and more efficient without making instruction loading stale.

How did you verify your code works?

• bun test test/session/message-v2.test.ts test/session/compaction.test.ts test/session/prompt-effect.test.ts test/session/processor-effect.test.ts
• bun typecheck
• Added prompt-loop regression coverage showing updated AGENTS.md instructions are visible on the next iteration.
• Added prompt-loop proof coverage showing SystemPrompt.skills() and SystemPrompt.environment() are only computed once across a two-iteration loop.

Screenshots / recordings

N/A

Checklist

• I have tested my changes locally
• I have not included unrelated changes in this PR

[github-actions]

The following comment was made by an LLM, it may be inaccurate:

Based on my search results, I found one potentially related PR:

Related PR:

• fix(opencode): wire OpenAI previous_response_id session caching #20848: fix(opencode): wire OpenAI previous_response_id session caching
  • This PR is explicitly mentioned in the current PR's description as related work. PR fix(opencode): preserve tool context through compaction and prompt loops #21425 builds on this functionality by threading OpenAI Responses tool follow-ups with previous_response_id for the store: true configuration case.

The current PR (#21425) appears to be the only PR addressing the specific combination of preserving tool context across compaction AND tool follow-ups. While there are other PRs related to compaction and tool handling (like #20289, #18764, #14393), none directly duplicate the functionality being added in PR #21425.

No duplicate PRs found

[github-actions]

This PR doesn't fully meet our contributing guidelines and PR template.

What needs to be fixed:

• PR description is missing required template sections. Please use the PR template.

Please edit this PR description to address the above within 2 hours, or it will be automatically closed.

If you believe this was flagged incorrectly, please let a maintainer know.

[github-actions]

This pull request has been automatically closed because it was not updated to meet our contributing guidelines within the 2-hour window.

Feel free to open a new pull request that follows our guidelines.