fix(feishu): render Markdown tables as v2 cards instead of degrading to text

[315zuicai]

Problem

Feishu IM post messages do not render Markdown tables — they appear blank or as raw | a | b | source characters on the client. Before this change, FeishuAdapter._build_outbound_payload reacted by force-degrading any table-bearing reply to plain text, which still surfaced the raw pipe/dash characters to the user.

This makes any agent that emits a markdown table (very common — comparison tables, status reports, structured data) look broken on Feishu.

Fix

Two parts:

A) gateway/platforms/feishu.py — when the outbound content contains a Markdown table, wrap it into a minimal v2 interactive card (schema: 2.0) whose body is one or more tag: markdown elements. Feishu's card markdown elements DO render tables correctly. If the card build raises for any reason, the existing plain-text fallback is preserved so the send pipeline cannot crash.

Two helpers are added:

• _split_markdown_for_card_elements() — splits long content into chunks that fit Feishu's per-element soft limit (~3400 chars) and per-card element cap (40), respecting fenced code blocks.
• _build_table_card_payload() — wraps the (chunked) markdown into the minimal card v2 envelope.

B) agent/prompt_builder.py — tightens the Feishu directive injected into the system prompt so the model knows which Markdown features Feishu IM actually renders. The previous wording (“Feishu renders Markdown … bold, italic, code blocks, and links are supported”) was misleading because it implied tables/headings work too. New wording explicitly lists what works (**bold**, *italic*, code, links, lists) and what doesn’t (tables, ATX headings, block quotes, hr), and recommends key-value bullet lists or bold lines as safe alternatives. This is best-effort steering — the code-side fix in (A) catches the cases where the model still emits a table.

Behaviour change

• Before: a reply containing a Markdown table → user sees raw | a | b | source.
• After: same reply → user sees a rendered interactive card with the table laid out properly.
• No-table replies are unchanged (regression-tested).
• Plain text is unchanged.

Tests

Adds 4 cases to TestFeishuAdapterMessaging in tests/gateway/test_feishu.py:

1. test_build_outbound_payload_wraps_table_into_v2_card — table input produces ("interactive", <v2 card with tag:markdown elements>).
2. test_build_outbound_payload_falls_back_to_text_when_card_build_fails — _build_table_card_payload raising → returns ("text", ...) instead of crashing.
3. test_build_outbound_payload_non_table_markdown_still_uses_post — non-table markdown still routes to the existing post path.
4. test_build_outbound_payload_plain_text_unchanged — plain text stays as text.

All 361 existing tests in test_feishu.py + test_feishu_bot_admission.py + test_stream_consumer.py continue to pass.

Validation

Verified live in a real Feishu IM session: a reply containing a 3-row × 4-col Markdown table previously rendered as raw source; with this change it renders as an interactive card with the table laid out correctly.

Diff size

• agent/prompt_builder.py: +13 / -2
• gateway/platforms/feishu.py: +124 / -3 (two new module-level helpers + a small change in _build_outbound_payload)
• tests/gateway/test_feishu.py: +78 / -0

[teknium1]

Thanks for addressing a real Feishu rendering regression. Current main still routes detected Markdown tables to raw text in plugins/platforms/feishu/adapter.py:4524-4534, so the underlying fix remains needed.

Problems

• The new return "interactive", card_json at gateway/platforms/feishu.py:4433 has no matching API-rejection fallback. Current main only retries as text when msg_type == "post" (plugins/platforms/feishu/adapter.py:1913-1936), and edit_message() has the same post-only guard (:1963-1971). A rejected card can therefore fail delivery rather than preserve the stated fallback.
• The adapter was moved to plugins/platforms/feishu/adapter.py by 560010547, so this needs a migration-aware salvage rather than a direct application.

Suggested changes

• Port the routing/helpers to the bundled plugin and add interactive rejection fallback coverage for both send() and edit_message().
• Update the tests to import the plugin adapter and exercise the send path in addition to payload construction.

Automated hermes-sweeper review.

[teknium1]

This changes the wire type to interactive, but the surrounding send and edit fallback logic is still guarded by msg_type == "post". Please add an interactive-card rejection fallback and a send-path test; otherwise a Feishu API rejection can fail delivery instead of degrading safely.