feat(intent): set BIP-322 PSBT_GLOBAL_GENERIC_SIGNED_MESSAGE (0x09)

[afsheenb]

Intent-proof PSBTs never set BIP-322's global 0x09 field (the encoded message being signed). Without it a co-signer cannot recompute the to_spend commitment from PSBT-internal data alone, making it impossible to reliably distinguish a genuine ownership proof from an ordinary fund-moving spend before contributing a partial signature.

Insert the 0x09 global field (type_value=0x09, key=empty) into the proof PSBT inside build_proof_psbt(), using the already-encoded message string that is computed there for the to_spend tagged hash. The field round-trips correctly through PSBT serialize/deserialize.

Add two unit tests: one verifying the field is set to the expected bytes, one verifying it survives a PSBT round-trip. Provides feature parity with arkade-os/ts-sdk#578.

Summary by CodeRabbit

• New Features
  • Proof-of-funds PSBTs now embed signed-message metadata to improve compatibility with message-verification workflows.
• Bug Fixes
  • Improved PSBT saving/loading so the embedded message data is preserved during serialize/deserialize.
• Tests
  • Added unit coverage to verify the embedded signed-message metadata is correctly included and round-trips reliably.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 81571209-13f0-47c7-ab42-ceb2c13be945

📥 Commits

Reviewing files that changed from the base of the PR and between a3f436c and 720e040.

📒 Files selected for processing (1)

• ark-core/src/intent.rs

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

• ark-core/src/intent.rs

---

Walkthrough

build_proof_psbt now inserts PSBT global unknown field 0x09 with the encoded IntentMessage bytes. Related intent documentation comments were updated, and new tests verify the field value and PSBT round-trip behavior using a Taproot-backed helper input.

Changes

BIP-322 PSBT Global Field

Layer / File(s)	Summary
Intent docs and serialization comments ark-core/src/intent.rs	Adds and adjusts documentation comments for Input, Intent, make_intent, and IntentMessage, including the message serialization description.
PSBT field insertion and tests ark-core/src/intent.rs	build_proof_psbt writes the encoded intent message into PSBT global unknown field 0x09; the test module adds a Taproot-backed dummy_input() helper and checks the field value and serialize/deserialize round-trip.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly matches the main change: adding the BIP-322 PSBT global signed message field to intent PSBTs.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
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 unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{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 `@ark-core/src/intent.rs`:
- Around line 636-647: The dummy_input fixture in intent::dummy_input is
inconsistent with its contract: the comment says it should be built from an
OP_TRUE leaf, but the leaf is created with ScriptBuf::new_op_return. Update
dummy_input to construct a trivially spendable tapscript that matches the stated
OP_TRUE behavior, or adjust the helper comment to reflect the actual leaf type
so tests don’t rely on an unspendable fixture.

🪄 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: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 0a97eb06-ca1a-4efc-9cf3-19b97705956f

📥 Commits

Reviewing files that changed from the base of the PR and between 4e8b696 and a3f436c.

📒 Files selected for processing (1)

• ark-core/src/intent.rs

[afsheenb]

@coderabbitai I've updated the docstrings and fixed the test snafu. All better?

[coderabbitai]

Tip

For best results, initiate chat on the files or code changes.

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
set -euo pipefail

echo "== Outline =="
ast-grep outline ark-core/src/intent.rs --view expanded || true

echo
echo "== Current diff for ark-core/src/intent.rs =="
git diff -- ark-core/src/intent.rs || true

echo
echo "== Relevant file slices =="
nl -ba ark-core/src/intent.rs | sed -n '1,260p'
echo
nl -ba ark-core/src/intent.rs | sed -n '260,520p'

Length of output: 3067

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

echo "== Public items and surrounding doc comments =="
cat -n ark-core/src/intent.rs | sed -n '30,210p'

echo
echo "== build_proof_psbt and IntentMessage area =="
cat -n ark-core/src/intent.rs | sed -n '300,520p'

echo
echo "== tests / dummy_input =="
cat -n ark-core/src/intent.rs | sed -n '642,780p'

Length of output: 21140

---

@afsheenb Yes — this looks addressed to me.

The public-facing docs I was worried about are now in place, and the test helper is consistent again: dummy_input() now builds from an OP_TRUE leaf and uses the matching control block/script pubkey path (Line 649, Line 660, Line 666 through Line 677).

No further concerns from my side on those points.

---

If you found this review helpful, would you consider giving us a shout-out on X?

Thank you for using CodeRabbit!

[arkana-ai-bot]

Reviewed ark-core/src/intent.rs at 720e040. Small, focused change; behavior is metadata-only (no consensus/signing effect).

Correctness of the 0x09 insertion

• psbt.unknown.insert(Key { type_value: 0x09, key: vec![] }, message.as_bytes().to_vec()) at ark-core/src/intent.rs:414-417 is placed after the per-input loop, so it runs exactly once. Good.
• The bytes match the JSON-encoded message string already used to build the to_spend tagged hash (message shadowed from message.encode()? at intent.rs:330-333), so the field and the commitment are consistent.
• In rust-bitcoin 0.32.7 (per ark-core/Cargo.toml:17), PSBT global key type 0x09 is unused, so it round-trips through unknown. The build_proof_psbt_0x09_survives_psbt_roundtrip test confirms this.

Cross-repo interop check

• ts-sdk sets the same field: packages/ts-sdk/src/intent/index.ts:160,314-330 (const PSBT_GLOBAL_GENERIC_SIGNED_MESSAGE = 0x09, empty key, UTF-8 message).
• arkd writes it too: pkg/ark-lib/intent/proof.go:158-163 (Unknowns = append(..., &psbt.Unknown{Key: []byte{0x09}, Value: []byte(message)})). btcsuite Unknown.Key is the raw key bytes (type byte only, no extra key data), which serializes to length-prefix + 0x09 — byte-identical to rust-bitcoin's Key { type_value: 0x09, key: vec![] }. Interop with the Go implementation is intact.

Server-side consumption — none, and that's fine

• arkd/pkg/ark-lib/intent/proof.go's Verify(proofB64, message, skip) takes the message as an out-of-band parameter (proof.go:57-87) and never reads the 0x09 field. So this is purely informational metadata for verifiers that don't already have the message; the actual security binding remains the to_spend commitment inside unsigned_tx.input[0].previous_output.txid. Not protocol-critical.

Naming nit

• "BIP-322 PSBT_GLOBAL_GENERIC_SIGNED_MESSAGE (0x09)" is not something BIP-322 actually defines; it's an Ark ecosystem convention shared across the three SDKs. The naming is copied consistently, so no action needed here — but if a future BIP allocates 0x09 as a real global type, older PSBTs will fail to deserialize on newer library versions. Same risk carried by ts-sdk and arkd; worth being aware of.

Tests

• Danger's "no test changes" warning is stale: two new tests are added (intent.rs:743-767) covering both the initial insertion and the serialize/deserialize round-trip. dummy_input() correctly builds a taproot input using G's x-coordinate as the internal key, a single OP_TRUE leaf at depth 0, and matching control block — sufficient for the metadata check. Only IntentMessage::Delete is exercised, but since 0x09 handling doesn't branch on the variant, that's acceptable.
• No test asserts that make_intent() (the public wrapper) also propagates the field. It does today because it doesn't mutate psbt.unknown, but a light regression guard there wouldn't hurt if you want defense in depth.

Doc comments

• The doc additions on Input, Input::new, Intent, Intent::new, Intent::serialize_proof, Intent::serialize_message, make_intent, and IntentMessage/IntentMessage::encode are accurate. No API surface changes.

No blockers. LGTM.