# rawcli — catalog & memory-hinting roadmap

The MVP is a list of **named transponders**, each surfaced through the **same UX**: a command
catalog + a datasheet memory map, both offered as **autocomplete suggestions in the completion
dropdown** — never on the bottom bar, never in command output.

## The UX contract (applies to every transponder)

- Completion dropdown offers: command names (by letter → `READ(`), opcodes (by hex/binary value →
  raw byte inserted, command as the label), and page/block landmarks with datasheet roles.
- **Must work in both hex AND binary entry modes**, rendering in the active base.
- Function-call args are base-10 (`READ(4)`); raw bytes are hex, `0b` to intermix binary.

## FIXED — raw byte-line autocomplete now fires in hex AND binary

Root cause: `entry._digit` applied each keystroke as `buf.document = Document(...)`, which resets
prompt_toolkit's `complete_state` and never re-fires completion — so the live menu never appeared
during raw byte entry. This hit **both** hex and binary (the earlier "hex works, binary doesn't"
was inaccurate; only command-name completion worked, because that path already used `insert_text`).
Fixed by appending the per-keystroke delta via `buf.insert_text` instead — `type_digit` only ever
appends (optionally after an auto-space) or drops, so the change is always a pure suffix, and
`insert_text` triggers completion exactly like an ordinary keystroke. Verified by driving the real
key handler: the menu now populates on every digit in both modes. Regression:
`TestKeyBindings::test_digit_binding_fires_completion_trigger_for_raw_bytes`.

The Ctrl-t / Ctrl-_ toggle correctly flips the breadcrumb `0x` <-> `0b` and requests a redraw
(verified, guarded by `test_toggle_updates_breadcrumb_prefix`). An end-to-end live-session test
(`TestRawcliLiveSession`) drives the real PromptSession via a pipe input. A human eyeball of the
rendered dropdown on a real terminal is the only thing left unautomated.

## Priority 1 — MIFARE Classic (do this first)

- **Identify**: distinguish 1K (SAK 0x08) / 4K (0x18) / Mini (0x09); today only named, no map.
- **Memory map** (block → role), mirroring the NTAG page map:
  - Block 0 = manufacturer block: UID + BCC + SAK + ATQA + manufacturer data.
  - Sector trailer (block 3,7,…,63 on 1K; last block of each sector on 4K): **Key A (0-5),
    access bits (6-8) + GPB (9), Key B (10-15)** — the access bits are the bit-level detail.
  - Other blocks = data block (with sector number).
  - 4K: sectors 0-31 are 4 blocks; sectors 32-39 are 16 blocks each.
- **Catalog**: READ/WRITE/HALT exist; note that reads/writes need prior AUTH (Key A/B) — the
  authenticate step is the reader's job before these.
- Autocomplete must work in hex + binary (see the bug above).

## After MFC — 15693 vicinity chips: ICODE SLIX ✅, ICODE SLIX2 ✅, NTAG5 (next)

1. **15693 header builder** ✅ — `iso15_frame()` (flags + command + NXP mfg 0x04), shared by all
   15693 catalogs. The **full request-flags byte** is settable: sub-carrier (0x01), data-rate,
   inventory, protocol-ext (0x08), and the mode-dependent bits 5-6 (Select/Address vs AFI byte +
   Nb_slots 1/16), plus option. `iso15_flags_decode` decodes every bit for the byte-0 hint.
2. **Per-chip identify** — `identify.name_iso15` distinguishes SLIX vs SLIX2 (READ_SIGNATURE 0xBD
   probe). **NTAG5 and ICODE DNA both carry UID `E0 04 01 18`** (identical type indicator — model
   comment confirms), so they can't be told apart by UID. Routing NTAG5 needs a **functional
   probe** (e.g. READ_CONFIG 0xC0 of session register 0xA0/0xA1 — NTAG5 has them, DNA doesn't).
   ⚠ HARDWARE-DEPENDENT: needs a brief NTAG5 tap to capture the discriminating response (and ideally
   a DNA capture to confirm it differs) before wiring — otherwise it risks misclassifying DNA.
3. **Catalogs** — `_SLIX`, `_SLIX2`, and **`_NTAG5`** ✅ all done, dispatched by `catalog_for`,
   verified live. `_NTAG5` = SLIX2 minus STAY_QUIET_PERSISTENT (NTAG5 rejects it) + READ_CONFIG
   (0xC0) / WRITE_CONFIG (0xC1) / PICK_RANDOM_UID (0xC2) / READ_SRAM (0xD2) / WRITE_SRAM (0xD3).
   **Follow-up:** the AES / ISO 29167-10 commands AUTHENTICATE (0x35) / CHALLENGE (0x39) / READBUFFER
   (0x3A) for Link/Boost (AES-capable) variants — a crypto sub-protocol with its own param format.
4. Autocomplete UX ✅ (hex + binary, command names + opcode-by-value; the byte-line completion bug
   is fixed).

`INVENTORY_READ` (0xA0) / `FAST_INVENTORY_READ` (0xA1) ✅ — in `_SLIX` (inherited by `_SLIX2`),
**full request body** via `_inv_read_frame`: `[AFI] | mask_len(bits) | mask value | first_block |
count-1`, with the inventory flags set by `iso15_frame(inventory=…, afi=…)`. Mask-less (default),
mask-based selective anticollision, and AFI filter all supported. **15693 is now datasheet-complete
end to end** (flags byte + both SLIX/SLIX2 command sets + inventory-read body).

**Hardware verify pending:** SLIX2 catalog was verified only against unit tests + a live-session
drive (no SLIX2 tag on the reader). Confirm on a real SLIX2 when one is available.

## Also queued

- **T5577 config block (block 0)** bitfield decode (modulation, bit-rate, block count, PWD/AOR) —
  currently just "config block".
- NTAG I2C plus (model exists, identify names it) — catalog + map.

See [[project_rawcli_mvp]] in memory for the named-transponder scope.