fix(mcp): return text asset content as UTF-8, not base64

[getappz]

Problem

MCP asset action=get base64-encoded all asset content unconditionally, regardless of mime type. Reading a text/markdown asset (e.g. a handoff) forced every consumer through a decode step — the asset get -> base64-decode dance — even though the bytes were already valid UTF-8.

Fix

The get handler now returns readable text for textual MIME types whose bytes are valid UTF-8, and reports which encoding it used:

• textual MIME (text/*, +json/+xml, application/{json,xml,javascript,typescript,toml,yaml}) and valid UTF-8 -> "content" as text with "encoding": "utf8"
• everything else (binary MIME, or invalid UTF-8) -> "encoding": "base64"

Callers branch on encoding instead of blindly decoding. Gating on MIME (not just UTF-8 validity) prevents binary assets that happen to be valid UTF-8 — e.g. an application/octet-stream of control bytes — from being mislabeled utf8. The over-inline-limit and read-error branches are unchanged.

Tests

• asset_get_returns_text_content_as_utf8_not_base64 — a text/markdown asset with a non-ASCII codepoint round-trips as encoding: "utf8" with readable content.
• asset_get_returns_base64_for_binary_with_valid_utf8_bytes — a valid-UTF-8 .bin (octet-stream) asset returns encoding: "base64".

Full cargo test --workspace green.

Compatibility

Breaking change to the tool's output contract: consumers that unconditionally base64-decode content now receive raw text for textual assets. They should switch to checking the encoding field.

Tracked as agentflare item 134 (child of the DX/tooling-friction bucket, item 118) — not a GitHub issue.

Summary by CodeRabbit

• New Features
  • Inline text assets are now returned as readable UTF-8 content when their type is recognized as textual and the bytes decode cleanly.
  • Inline responses now include an encoding indicator (utf8 for text, base64 otherwise).
• Bug Fixes
  • Inline handling no longer forces Base64 for text content; non-textual or undecodable bytes are still Base64-encoded.
• Tests
  • Added unit tests covering UTF-8 vs Base64 behavior for inline assets based on both decode results and asset type.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: c6dd74f2-ee97-4799-b09d-4b3b0ad33002

📥 Commits

Reviewing files that changed from the base of the PR and between 6fa528c and 01dd2e2.

📒 Files selected for processing (1)

• src/mcp_server.rs

🚧 Files skipped from review as they are similar to previous changes (1)

• src/mcp_server.rs

---

📝 Walkthrough

Walkthrough

Inline asset retrieval now returns readable UTF-8 for decodable textual assets, reports the encoding, and falls back to Base64 for binary or invalid UTF-8 content. Tests cover both encoding outcomes.

Changes

Inline asset encoding

Layer / File(s)	Summary
Encoding selection and validation src/mcp_server.rs	A MIME classifier identifies textual assets; inline retrieval returns original text with "encoding": "utf8" only when bytes are valid UTF-8 and textual, otherwise Base64 with "encoding": "base64". Tests cover markdown text and binary content containing valid UTF-8 bytes.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• getappz/agentflare#168: Both changes modify inline asset encoding in the MCP asset tool.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly matches the main change: returning textual asset content as UTF-8 instead of base64.
Description check	✅ Passed	The description covers the problem, fix, tests, and compatibility, though it doesn't use the exact template headings.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/asset-get-utf8-content

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@src/mcp_server.rs`:
- Around line 2745-2748: Update the content/encoding selection around
std::str::from_utf8 to first allow text decoding only for MIME types in the
established textual MIME allowlist; return Base64 for binary MIME types even
when their bytes are valid UTF-8. Add a regression test covering a valid-UTF8
.bin asset containing binary bytes and verify it reports "encoding": "base64".

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 32e030eb-905b-4766-ad3e-fa97b2671b15

📥 Commits

Reviewing files that changed from the base of the PR and between e9fdea9 and 0f6ea2c.

📒 Files selected for processing (1)

• src/mcp_server.rs

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@src/mcp_server.rs`:
- Around line 5138-5152: Update the test around the parsed `got` response to
assert that `got["content"]` equals the Base64 string `AAECAw==` for the binary
input `[0, 1, 2, 3]`, while preserving the existing `encoding` assertion.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 8f06321c-0142-4da8-911a-b58652a164a3

📥 Commits

Reviewing files that changed from the base of the PR and between 0f6ea2c and 7cd0ce5.

📒 Files selected for processing (1)

• src/mcp_server.rs